Microsoft Word - NCAP-ADAS-RFC-03-03-2022-web.docx

The Deputy Administrator of the National Highway Traffic Safety Administration, Steven S. Cliff,

Ph.D., signed the following Request for Comment (RFC) on March 3, 2022, which the Agency is

submitting for publication in the Federal Register. While NHTSA has taken steps to ensure the

accuracy of this Internet version of the RFC, it is not the official version of the RFC. Please refer

to the official version in a forthcoming Federal Register publication, which will appear on the

Government Printing Office’s FDSys website (www.gpo.gov/fdsys/search/home.action) and on

Regulations.gov (http://www.regulations.gov/). Once the official version of this document is

published in the Federal Register, this version will be removed from the Internet and replaced

with a link to the official version.

DEPARTMENT OF TRANSPORTATION

National Highway Traffic Safety Administration

[Docket No. NHTSA-2021-0002]

New Car Assessment Program

AGENCY: National Highway Traffic Safety Administration (NHTSA), Department of

Transportation (DOT).

ACTION: Request for comments (RFC).

SUMMARY: NHTSA’s New Car Assessment Program (NCAP) provides comparative

information on the safety performance of new vehicles to assist consumers with vehicle

purchasing decisions and to encourage safety improvements. In addition to star ratings for crash

protection and rollover resistance, the NCAP program recommends particular advanced driver

assistance systems (ADAS) technologies and identifies the vehicles in the marketplace that offer

the systems that pass NCAP performance test criteria for those systems. This notice proposes

significant upgrades to NCAP, first, by proposing to add four more ADAS technologies to those

NHTSA currently recommends. The new technologies are blind spot detection, blind spot

intervention, lane keeping support, and pedestrian automatic emergency braking. This notice

also proposes changes (including an increase in stringency) to the test procedures and

performance criteria for the four currently recommended ADAS technologies in NCAP to enable

enhanced evaluation of their capabilities in current vehicle models and to harmonize with other

consumer information programs. Second, this notice describes (but does not propose at this

time) how NHTSA could rate vehicles equipped with these ADAS technologies and requests

comment on how best to develop this rating system. Third, NHTSA seeks (but does not propose

at this time) to provide a crash avoidance rating at the point of sale on a vehicle’s window

sticker, consistent with the 2015 Fixing America’s Surface Transportation (FAST) Act, and

discusses ways of implementing the program, including a potential process for updating such

information. Fourth, as part of a new NHTSA approach to NCAP, NHTSA is proposing a

“roadmap” of the Agency’s plans to upgrade NCAP in phases over the next several years and

presents the roadmap for comment. Fifth, as another first for NCAP, NHTSA is considering

utilizing NCAP to raise consumer awareness of certain safety technologies that may have the

potential to help people make safe driving choices. This information may be of particular

interest to parents or other caregivers shopping for a vehicle for a new or inexperienced driver in

the household, or parents wanting to know more about rear seat alerts for hot

car/heatstroke. Sixth and finally, this RFC discusses NHTSA’s ideas for updating several

programmatic aspects of NCAP to improve the program. The proposal on ADAS technologies

and the aforementioned initiatives pave the way for the Agency to focus on a much broader

safety strategy, including fulfilling not only the 2015 FAST Act directive but also the recent

mandates included in Section 24213 of the November 2021 Bipartisan Infrastructure Law,

enacted as the Infrastructure Investment and Jobs Act, to improve road safety for motor vehicle

occupants as well as other vulnerable road users.

DATES: Comments should be submitted no later than [INSERT DATE 60 DAYS AFTER

DATE OF PUBLICATION IN THE FEDERAL REGISTER].

ADDRESSES: Comments should refer to the docket number above and be submitted by one of

the following methods:

 Federal Rulemaking Portal: http://www.regulations.gov. Follow the online instructions

for submitting comments.

 Mail: Docket Management Facility, U.S. Department of Transportation, 1200 New Jersey

Avenue S.E., West Building Ground Floor, Room W12-140, Washington, D.C. 20590-

0001.

 Hand Delivery: 1200 New Jersey Avenue S.E., West Building Ground Floor, Room

W12-140, Washington, D.C., between 9 a.m. and 5 p.m. ET, Monday through Friday,

except Federal Holidays.

 Instructions: For detailed instructions on submitting comments, see the Public

Participation heading of the SUPPLEMENTARY INFORMATION section of this

document. Note that all comments received will be posted without change to

http://www.regulations.gov, including any personal information provided.

 Privacy Act: Anyone can search the electronic form of all comments received in any of

our dockets by the name of the individual submitting the comment (or signing the

comment, if submitted on behalf of an association, business, labor union, etc.). You may

review DOT's complete Privacy Act Statement in the Federal Register published on

April 11, 2000 (65 FR 19477-78) or at https://www.transportation.gov/privacy. For

access to the docket to read background documents or comments received, go to

http://www.regulations.gov or the street address listed above. Follow the online

instructions for accessing the dockets.

FOR FURTHER INFORMATION CONTACT: For technical issues, you may contact Ms.

Jennifer N. Dang, Division Chief, New Car Assessment Program, Office of Crashworthiness

Standards (Telephone: 202-366-1810). For legal issues, you may contact Ms. Sara R. Bennett,

Office of Chief Counsel (Telephone: 202-366-2992). You may send mail to either of these

officials at the National Highway Traffic Safety Administration, 1200 New Jersey Avenue S.E.,

West Building, Washington, DC 20590-0001.

SUPPLEMENTARY INFORMATION:

Table of Contents

I. Executive Summary

II. Background

III. ADAS Performance Testing Program

A. Lane Keeping Technologies

1. Updating Lane Departure Warning (LDW)

a. Haptic Alerts

b. False Positive Tests

c. LDW Test Procedure Modifications

2. Adding Lane Keeping Support (LKS)

B. Blind Spot Detection Technologies

1. Adding Blind Spot Warning (BSW)

a. Additional Test Targets and/or Test Conditions

b. Test Procedure Harmonization

2. Adding Blind Spot Intervention (BSI)

C. Adding Pedestrian Automatic Emergency Braking (PAEB)

D. Updating Forward Collision Prevention Technologies

1. Forward Collision Warning (FCW)

2. Automatic Emergency Braking (AEB)

a. Dynamic Brake Support (DBS)

b. Crash Imminent Braking (CIB)

c. Current State of AEB Technology

d. NHTSA’s CIB Characterization Study

e. Updates to NCAP’s CIB Testing

f. Updates to NCAP’s DBS Testing

g. Updates to NCAP’s FCW Testing

h. Regenerative Braking

3. FCW and AEB Comments Received in Response to 2015 RFC Notice

a. Forward Collision Warning (FCW) Effective Time-to-Collision

b. False Positive Test Scenarios

c. Procedure Clarifications

d. Expand Testing

e. AEB Strikeable Target

IV. ADAS Rating System

A. Communicating ADAS Ratings to Consumers

1. Star Rating System

2. Medals Rating System

3. Points-Based Rating System

4. Incorporating Baseline Risk

B. ADAS Rating System Concepts

1. ADAS Test Procedure Structure and Nomenclature

2. Percentage of Test Conditions to Meet – Concept 1

3. Select Test Conditions to Meet – Concept 2

4. Weighting Test Conditions Based on Real-World Data – Concept 3

5. Overall Rating

V. Revising the Monroney Label (Window Sticker)

VI. Establishing a Roadmap for NCAP

VII. Adding Emerging Vehicle Technologies for Safe Driving Choices

A. Driver Monitoring Systems

B. Driver Distraction

C. Alcohol Detection

D. Seat Belt Interlocks

E. Intelligent Speed Assist

F. Rear Seat Child Reminder Assist

VIII. Revising the 5-Star Safety Rating System

A. Points-Based Ratings System Concept

B. Baseline Risk Concept

C. Half-Star Ratings

D. Decimal Ratings

E. Rollover Resistance Test

IX. Other Activities

A. Programmatic Challenges with Self-Reported Data

B. Website Updates

C. Database Changes

X. Economic Analysis

XI. Public Participation

XII. Appendices

I. Executive Summary

NHTSA’s New Car Assessment Program (NCAP) supports NHTSA’s mission to

reduce the number of fatalities and injuries that occur on U.S. roadways. NCAP, like

many other NHTSA programs, has contributed to significant reductions in motor vehicle

fatalities. In the decade prior to the 1978 start of NCAP, fatalities from motor vehicle

crashes exceeded 50,000 annually. In 2019, 36,096 people still lost their lives on U.S.

roads. Passenger vehicle occupant fatalities decreased from 32,225 in 2000 to 22,215 in

2019.

This reduction is notable, particularly in light of the fact that the total number of

vehicle miles traveled (VMT) in the U.S. has increased over time. However, during that

same timeframe, pedestrian fatalities increased by 33 percent, from 4,739 in 2000 to

Traffic Safety Facts 2019 “A Compilation of Motor Vehicle Crash Data.” U.S. Department of Transportation.

National Highway Traffic Safety Administration.

6,205 in 2019.

Furthermore, a statistical projection of traffic fatalities for the first half of 2021

shows that an estimated 20,160 people died in motor vehicle traffic crashes – the highest number

of fatalities during the first half of the year since 2006, and the highest half-year percentage

increase in the history of data recorded by the Fatality Analysis Reporting System (FARS).

addition, the projected 11,225 fatalities during the second quarter of 2021 represents the highest

second quarter fatalities since 1990, and the highest quarterly percentage change (+23.1 percent)

in FARS data recorded history. Preliminary data reported by the Federal Highway

Administration (FHWA) show that VMT in the first half of 2021 rebounded from a large

pandemic-related dip that occurred in the first half of 2020, increasing by 173.1 billion miles, or

about a 13 percent increase over the comparable period in 2020. The fatality rate for the first

half of 2021 increased to 1.34 fatalities per 100 million VMT, up from the projected rate of 1.28

fatalities per 100 million VMT in the first half of 2020. Early evidence suggests that these

fatality rates have increased as a result of increases in risky behaviors like driving and riding

while unbelted, impaired driving, and speeding.

Although there have been notable gains in

automotive safety over the past fifty years, far more work must be done.

This notice discusses how NCAP can support NHTSA’s mission through its multi-

faceted initiatives and broad safety strategies to address vehicle safety involving motor vehicle

occupants, other vulnerable road users, and safe driving choices to further reduce injuries and

fatalities occurring on the nation’s roads. As stated in the Department of Transportation’s

National Roadway Safety Strategy, proposals to update NCAP are expected to emphasize safety

Traffic Safety Facts 2000 “A Compilation of Motor Vehicle Crash Data from the Fatality Analysis Reporting

System and the General Estimates System.” U.S. Department of Transportation. National Highway Traffic Safety

Administration.

National Center for Statistics and Analysis. (2021, October), Early Estimate of Motor Vehicle Traffic Fatalities for

the First Half (January-June) of 2021. (Traffic Safety Facts. Report No. DOT HS 813 199), Washington, DC:

National Highway Traffic Safety Administration.

See https://www.nhtsa.gov/press-releases/2020-fatality-data-show-increased-traffic-fatalities-during-pandemic.

features that protect people both inside and outside of the vehicle, and may include

consideration of pedestrian protection systems, better understanding of impacts to

pedestrians (e.g., specific

considerations for children), and automatic emergency braking and lane keeping assistance to

benefit bicyclists and pedestrians. In a first-of-its-kind focus – especially relevant in light of

increases in fatalities caused by risky driving behaviors – this notice seeks comment on how

automakers could encourage consumers to choose safety technologies that could prevent risky

behaviors from occurring in the first place. This notice also proposes significant upgrades to

NCAP by adding four additional crash avoidance technologies (also termed ADAS throughout

this notice) to the program, increasing the stringency of the tests for currently recommended

ADAS technologies in NCAP for enhanced evaluation of their current capabilities, and

exploring, for the first time, expanding NCAP to include safety for road users outside of the

vehicle. Finally, this document presents a roadmap of NHTSA’s current plans to upgrade NCAP

in phases over the next several years.

Many of these efforts align with Section 24213 of the Bipartisan Infrastructure

Law, enacted as the Infrastructure Investment and Jobs Act

and signed on November 15,

2021. First, this RFC, once finalized, fulfills the requirements of Section 24213(a) of the

Bipartisan Infrastructure Law because NHTSA intends for the addition of the four

technologies proposed in this RFC to “finalize the proceeding for which comments were

requested” on December 16, 2015.

Specifically, the finalization of this RFC will close

the December 16, 2015 proceeding and notice. While NHTSA has future plans described

in the roadmap that the Agency discussed in the December 16, 2015 notice, none are

(Pub. L. 117-58).

Id. at Section 24213(a); the notice referred to in the Bipartisan Infrastructure Law is 80 FR 78522 (Dec. 16, 2015).

This is the notice that will be finalized once the final decision notice for today’s RFC is published.

considered an extension of the December 16, 2015 proceeding, though all information previously

collected by NHTSA may be used in the development of future notices.

Second, this RFC fulfills portions of the requirements in Section 24213(b) of the

Bipartisan Infrastructure Law that mandates the Agency “publish a notice, for the

purposes of public comment, to establish a means for providing consumer information

relating to advanced crash-avoidance technologies” within one year of enactment that

includes: (1) an appropriate methodology for determining which advanced crash avoidance

technologies should be included in the information, (2) performance test criteria for use by

manufacturers in evaluating those technologies, (3) a distinct rating system involving each

technology, and (4) updating overall vehicle ratings to include the new rating. Through this

RFC, NHTSA is proposing four additional advanced crash avoidance technologies

for inclusion

in NCAP, proposing the test criteria for evaluating the advanced crash avoidance technologies,

and seeking comment on the future development of a crash avoidance rating system. NHTSA

described in detail why it chose the four technologies that it did and how those technologies meet

NHTSA’s established criteria for inclusion in NCAP. Since NHTSA is proposing the addition of

four advanced crash avoidance technologies and test criteria for evaluating those technologies,

NHTSA meets two of the four requirements for fulfillment of the Advanced Crash Avoidance

section of Sec. 24213(b).

Section 24213(b) of the law also requires that the Agency publish a notice “to establish a

means for providing to consumers information relating to pedestrian, bicyclist, or other

vulnerable road user safety technologies” within one year of enactment. This notice must meet

requirements very similar to the advanced crash avoidance notice mentioned above. Since

NHTSA is today proposing to include pedestrian automatic emergency braking (PAEB) in the

This notice refers to the advanced crash avoidance technologies as Advanced Driver Assistance Systems (ADAS)

technologies.

program and is including test criteria for evaluating PAEB, NHTSA meets two of the

four requirements for fulfillment of the Vulnerable Road User Safety section of Sec.

24213(b). The remaining requirements will be fulfilled once NHTSA proposes and then

finalizes a new rating system for the crash avoidance technologies in NCAP. The law

also requires that NHTSA submit reports to Congress on its plans for fulfilling the

abovementioned requirements. NHTSA plans to fulfill these reporting requirements in a

timely manner.

Third, this RFC, once finalized, fulfills the requirements of Section 24213(c) for

NHTSA to establish a roadmap for implementation of NCAP changes that covers a term

of ten years, with five year mid-term and five year long-term components, and with

updates to the roadmap at least once every four years to reflect new Agency interests and

public comments. The first roadmap must be completed within one year of the law’s

enactment. Once finalized, the roadmap on future updates to NCAP proposed in this

RFC in its entirety would fulfill the ten-year roadmap requirement, as some proposed

initiatives will be considered in NCAP in the first five years while others will be

proposed in the second half of the ten-year plan. The details and analysis of this

fulfillment are available in the Roadmap section of this RFC.

Fourth, this RFC, once finalized, will fulfill a provision in Section 24213(c) of the

Bipartisan Infrastructure Law that requires NHTSA to make the roadmap available for

public comment and to consider the public comments received before finalizing the

roadmap. These provisions are in accordance with the Agency’s current practice for

updating NCAP and will be followed to finalize the roadmap. Section 24213(c) of the

Law also requires that NHTSA identify opportunities where NCAP would “benefit from

harmonization with third-party safety rating programs.” The Agency is taking steps to

harmonize with existing consumer information rating programs where possible, and when

appropriate, as noted in various sections of this RFC.

Fifth, Section 24213(c) of the Law requires the Agency to engage with

stakeholders with diverse backgrounds and viewpoints not less than annually to develop future

roadmaps. Again, this provision is in accordance with the Agency’s current practice.

Components of the Notice

There are six main parts to this notice:

1. Proposes to add four new ADAS technologies to NCAP and updates to current NCAP

test procedures,

2. Discusses the Agency’s plan to develop a new rating system for advanced driver

assistance technologies,

3. Describes steps to list the crash avoidance rating information on the vehicle’s window

sticker (the Monroney label) at the point of sale,

4. Describes roadmap of the Agency’s plans to update NCAP in phases over the next ten

years,

5. Requests comments on expanding NCAP to provide consumer information on safety

technologies that could help people drive safer by preventing or limiting risky driving

behavior, and

6. Discusses NHTSA’s ideas for updating several programmatic aspects of NCAP to improve

the program as a whole.

Each of the aforementioned aspects of the notice are described in greater detail that

follows. First, the notice discusses in detail the Agency’s proposed upgrade to add four more

ADAS technologies to those currently recommended by NHTSA through NCAP and that are

highlighted on the NHTSA website. Since 2010, NCAP has recommended four kinds of ADAS

technologies to prospective vehicle purchasers, and has identified to shoppers the vehicles that

have these technologies and that meet NCAP performance test criteria.

The current

technologies are forward collision warning (FCW), lane departure warning (LDW), crash

imminent braking (CIB), and dynamic brake support (DBS) (with the latter two collectively

referred to as “automatic emergency braking).

This notice proposes changes (including

an increase in stringency) to the test procedures and performance criteria for LDW, CIB,

DBS, and FCW to (1) enable enhanced evaluation of their capabilities in current vehicle

models, (2) reduce test burden, and (3) harmonize with other consumer information

programs. This notice also describes and proposes four more ADAS technologies: blind

spot detection, blind spot intervention, lane keeping support, and pedestrian automatic

emergency braking.

These four new ADAS technologies are candidates for NCAP because data indicate they

satisfy NHTSA’s four prerequisites for inclusion in the program. The prerequisites are: (1) the

update to the program addresses a safety need; (2) there are system designs (countermeasures)

that can mitigate the safety problem; (3) existing or new system designs have safety benefit

potential; and (4) a performance-based objective test procedure exists that can assess system

performance. In order to address (1), a safety need, the Agency inherently looks first to address

injuries and fatalities stemming from “high-frequency and high-risk crash types” – as these

crashes command the largest safety need and thus may also afford the biggest potential benefit.

NHTSA does not calculate relative costs and benefits when considering inclusion in NCAP as it

is a non-regulatory consumer information program. NHTSA discusses in this notice how each of

the proposed ADAS technologies meets the four prerequisites. As explained in detail in this

NCAP only indicates that a vehicle has a recommended technology when NHTSA has data verifying that the

technology meets the minimum performance requirements set by NHTSA for acceptable performance. If a vehicle’s

ADAS is reported to have satisfied the performance requirements using the test methods specified by the Agency,

then NHTSA uses a checkmark system to indicate on the NHTSA website that the vehicle is equipped with the

technology. Each year, NHTSA also selects a sample of vehicles from that model year to verify ADAS system

performance by performing its own tests.

https://www.nhtsa.gov/equipment/driver-assistance-technologies.

notice, the four new ADAS technologies proposed in NCAP are the only technologies that the

Agency believes meet the four prerequisites for inclusion at this time. Each technology has

demonstrated the ability to successfully mitigate high frequency and high-risk crash types. With

the proposal to include pedestrian automatic emergency braking, NCAP would be expanded, for

the first time, to include safety for people outside of the vehicle.

Second, this notice discusses the Agency’s plan to develop a future rating system

for new vehicles based on the availability and performance of all the NCAP-recommended crash

avoidance technologies. Currently, NCAP only recommends crash avoidance technologies to

shoppers, and identifies the vehicles that offer the recommended technologies that pass NCAP

system performance criteria. Unlike its crashworthiness and rollover protection programs that

offer a combined rating based on vehicle performance in frontal, side, and rollover tests, the

NCAP crash avoidance program does not currently have a rating system to differentiate the

performance of ADAS technologies. NHTSA seeks to remedy this by developing a rating

system for ADAS technologies to provide purchasers improved data with which to compare and

shop for vehicles, and to spur improved vehicle performance. Accordingly, this document seeks

public input on how best to develop this rating system.

Third, this notice announces NHTSA’s steps to list the crash avoidance rating

information on the vehicle’s window sticker (the Monroney label) at the point of sale, as directed

by the FAST Act.

NHTSA requests comment on ideas for the Monroney label information.

Research is underway to maximize the effectiveness of the information in informing purchasing

decisions. A follow-on notice will propose the crash avoidance rating system and explain how

NHTSA would use the ratings. NHTSA will consider the comments received on this notice in

conjunction with the information gained from the consumer research, to develop a proposal for a

This Act requires NHTSA to promulgate a rule to require vehicle manufacturers to include crash avoidance

information next to the crashworthiness information on vehicle window stickers (Monroney labels).

revised label. To help shoppers make more informed purchasing decisions, NHTSA also

plans to provide fuel economy and greenhouse gas rating information with the NHTSA

safety ratings, not only at the point of sale but also on the NHTSA website.

Fourth, as part of a new approach to advancing NCAP, NHTSA has developed a roadmap

of the Agency’s current plans to upgrade NCAP in phases over the next several years.

The roadmap sets forth NHTSA’s near-term and longer-term strategies for upgrading

NCAP. The roadmap takes a gradual approach, which contemplates NHTSA’s issuing

proposed upgrades in phases, as the technologies mature to readiness for proposed

inclusion in NCAP. Following a proposal will be a final decision document that responds

to comments and provides NHTSA’s decisions for that phase of NCAP updates,

including the lead time provided for the implementation. The roadmap presents an

estimated timeframe of the phased request for comment (RFC) notices.

Fifth, this notice also considers expanding NCAP to provide consumer

information on safety technologies that could help people drive safer by preventing or

limiting risky driving behavior. The Agency is examining the possibility of expanding

NCAP to include technologies that promote NHTSA’s continuing efforts to combat

unsafe driving behaviors, such as distracted and impaired driving, riding in a vehicle

unrestrained, and speeding. NHTSA currently uses many approaches to reduce

dangerous driving behaviors, including high visibility enforcement and advertising

campaigns like “Click it or Ticket” and “Buzzed Driving is Drunk Driving.” These

campaigns have succeeded in reducing, but not eliminating, human causes of crashes and

there is some evidence that their success has reached a plateau. NHTSA is considering

how NCAP can promote technologies that would reduce unsafe driving or riding

behavior like distracted and impaired driving, speeding, or riding in a vehicle

unrestrained by targeting the human behaviors most likely to lead to crashes. This

information may be of particular interest to parents or other caregivers who are shopping for a

vehicle for a new or inexperienced driver in the household, or caregivers wanting to know more

about rear seat alerts for hot car/heatstroke.

Sixth and finally, this RFC discusses NHTSA’s ideas for updating several programmatic

aspects of NCAP to improve the program as a whole. NHTSA requests comment on the

Agency’s ideas for revising the 5-star safety ratings program. This document also discusses

ways NHTSA would like to update the existing ADAS technology program components,

outlines challenges the Agency has encountered relating to manufacturer self-reported data, and

proposes possible solutions to those problems. Lastly, the RFC discusses 1) updates to the

NCAP website to improve the dissemination of vehicle safety information to consumers and 2)

the development of an NCAP database to modernize the operational aspects of the program,

including a new vehicle information submission process for vehicle manufacturers.

This RFC includes numbered questions throughout the notice that highlight specific

topics on which NHTSA seeks comments. Although several questions may be posed un-

numbered within the body of certain sections, these un-numbered questions are reiterated at the

conclusion of the topic discussion and in Appendix B. To help ensure that NHTSA is able to

address all comments received, the Agency requests that commenters provide corresponding

numbering in their responses.

II. Background

NHTSA established its NCAP in 1978 in response to Title II of the Motor Vehicle

Information and Cost Savings Act of 1972. When the program first began providing consumers

with vehicle safety information derived from frontal crashworthiness testing, attention within the

industry to vehicle safety was relatively new. Today’s consumers are much more interested in

vehicle safety, and this has become one of the key factors in vehicle purchasing

decisions.

Vehicle manufacturers have responded to these consumer demands by

offering safer vehicles that incorporate enhanced safety features. This has resulted in

improved vehicle safety performance in NCAP, which has historically translated into

higher NCAP star ratings.

Over the years, NHTSA began to incorporate ADAS technologies into NCAP’s

crash avoidance program. In 2007, NHTSA, for the first time, issued an RFC exploring

the addition of ADAS technologies in NCAP.

Later, based on feedback received from

written and oral comments, NHTSA published a final decision

expanding NCAP to

include certain ADAS technologies and specific performance thresholds that a NHTSA-

recommended ADAS system must meet. Beginning with model year 2011, the Agency

began recommending on its website forward collision warning (FCW), lane departure

warning (LDW), and electronic stability control (ESC),

and identified to shoppers

which vehicles have the technologies that meet NCAP’s performance requirements.

NHTSA updated NCAP further to include crash imminent braking (CIB) and dynamic

braking support (DBS) technologies, beginning with model year 2018 vehicles.

This RFC continues those efforts. Through several notices and public meetings,

NHTSA has continued discussions with stakeholders about which technologies should be

included in NCAP and the minimum performance thresholds those technologies should

See www.regulations.gov, See www.regulations.gov, Docket No. NHTSA-2020-0016 for a report of “New Car

Assessment Program 5-Star Quantitative Consumer Research.”

72 FR 3473 (January 25, 2007). The RFC included a request for comments on a NHTSA report titled, “The New

Car Assessment Program (NCAP); Suggested Approaches for Future Enhancements.”

73 FR 40016 (July 11, 2008).

ESC was removed from the Agency’s list of recommended ADAS technologies through NCAP beginning in

model year 2014 when the technology became mandated under FMVSS No. 126, “Electronic stability control.”

NHTSA also included rear video systems in its list of recommended technologies under NCAP from model years

2014 to 2017 and removed that technology from its list when it became mandated under FMVSS No. 111, “Rear

Visibility.”

meet. NHTSA has set forth in Appendix C to this RFC a detailed history of the requests for

comment, public meetings, and other relevant events that underlie this notice.

The last RFC NHTSA published to discuss potential changes to NCAP was published in

2015. It was broad in subject matter and sought comment on NCAP’s potential use of enhanced

tools and techniques for evaluating the safety of vehicles, generating star ratings, and stimulating

further vehicle safety developments.

On the crashworthiness front, the RFC sought comment

on establishing a new frontal oblique test and on using more advanced crash test dummies in all

tests. The RFC also sought comment about establishing a new crash avoidance rating category

and including nine advanced crash avoidance technologies. Additionally, the RFC sought

comment on establishing a new pedestrian protection rating category involving the use of adult

and child head, upper leg, and lower leg impact tests and adding two new pedestrian crash

avoidance technologies. The RFC sought comment on combining the three categories (crash

avoidance, crashworthiness, and pedestrian protection) into one overall 5-star rating. NHTSA

also received comments at two public hearings, one in Detroit, Michigan, on January 14, 2016,

and the second at the U.S. DOT Headquarters in Washington, D.C., on January 29, 2016. The

numerous comments received on the RFC are discussed in a section below.

In October 2018, NHTSA hosted a third public meeting to re-engage stakeholders and

seek up-to-date input to help the Agency plan the future of NCAP.

The Agency has also been

working to finalize its research efforts on pedestrian crash protection, advanced anthropomorphic

test devices (crash test dummies) in frontal and side impact tests, a new frontal oblique crash test,

and an updated rollover risk curve. As discussed in the roadmap, NHTSA plans to upgrade the

NCAP crashworthiness program in phases over the next several years with the knowledge it has

acquired from the research programs.

80 FR 78521 (Dec. 16, 2015).

October 1, 2018.

III. ADAS Performance Testing Program

ADAS technologies have the potential to increase safety by preventing crashes or

mitigating the severity of crashes that might otherwise lead to injury and death. NCAP currently

conducts performance verification tests for four ADAS technologies: forward collision warning

(FCW), lane departure warning (LDW), crash imminent braking (CIB), and dynamic brake

support (DBS). CIB and DBS are collectively referred to as automatic emergency braking

(AEB). Vehicles that are equipped with one or more of these systems and pass NCAP’s

performance test requirements are listed as “Recommended” on NHTSA’s website. When the

Agency first began recommending FCW and LDW systems for model year 2011 vehicles, the

fitment rate for these systems was less than 0.2 percent (where “fitment rate” means the percent

of vehicles equipped with a particular ADAS system). For model year 2018 vehicles, 38.3

percent were equipped with FCW and 30.1 percent were equipped with LDW.

Providing

vehicle safety information through NCAP can be an effective approach to advance the

deployment of safer vehicle designs and technology in the U.S. market, inform consumer

choices, and encourage adoption of new technologies that have life-saving potential.

With this notice, NHTSA is proposing to incorporate four additional ADAS technologies

into NCAP’s crash avoidance program: lane keeping support (LKS), pedestrian automatic

emergency braking (PAEB), blind spot warning (BSW), and blind spot intervention (BSI). Each

of these technologies meets the Agency’s established criteria for inclusion in NCAP: (1) the

technology addresses a safety need; (2) system designs exist that can mitigate the safety problem;

(3) the technology provides the potential for safety benefits; and (4) a performance-based

objective test procedure exists that can assess system performance.

Details about how each of

the proposed ADAS technologies addresses a safety need (criterion 1) will be discussed

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

78 FR 20599 (Apr. 5, 2013).

immediately below, while the remaining criteria will be discussed in the relevant sections under

each technology.

To gain an understanding of the safety need that current ADAS technologies may

address, NHTSA analyzed crash data for 84 mutually exclusive pre-crash scenarios.

The pre-

crash scenarios used in the Agency’s analysis were devised using a typology

concept

published by the Volpe National Transportation Systems Center (Volpe),

which categorizes

crashes into dynamically distinct scenarios based on pre-crash vehicle movements and critical

events. As detailed in the referenced March 2019 report, NHTSA mapped the pre-crash scenario

typologies to twelve currently available ADAS technologies

believed to potentially address

certain pre-crash scenarios by assisting the driver to avoid or mitigate a crash. These mappings

served to define the corresponding crash populations (i.e., target crash populations).

Since several ADAS technologies presently available on passenger vehicles

are

designed to mitigate the same crash scenarios, NHTSA first grouped the technologies with

similar design intent into categories. The five technology categories that resulted from this

grouping process include: (1) forward collision prevention, (2) lane keeping, (3) blind spot

detection, (4) forward pedestrian impact, and (5) backing collision avoidance. As shown in

Table A-6, these categories address the following high-level crash types: (1) rear-end; (2)

rollover, lane departure, and road departure; (3) lane change/merge; (4) pedestrian; and (5)

backing, respectively. Of the original 84 pre-crash scenarios studied, we mapped 34 relevant

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

A typology is the study or analysis of something, or the classification of something, based on types or categories.

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019), Statistics of light-vehicle pre-

crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington, DC: National

Highway Traffic Safety Administration.

The twelve ADAS technologies were as follows: FCW, DBS, CIB, LDW, LKS, lane centering assist (LCA),

BSW, BSI, lane change/merge warning, PAEB, RAB, and rear cross-traffic alert.

Passenger vehicles were defined as cars, crossovers, sport utility vehicles (SUVs), light trucks, and vans having a

gross vehicle weight rating (GVWR) of 10,000 pounds or less.

pre-crash scenario typologies to the five resulting technology categories that represented these

crash types.

The forward collision prevention category included three ADAS technologies: forward

collision warning, crash imminent braking, and dynamic brake support (FCW, CIB, and DBS,

respectively). The lane keeping category included lane departure warning (LDW), lane keeping

support (LKS),

and lane centering assist (LCA). The blind spot detection category included

blind spot warning (BSW),

blind spot intervention (BSI), and lane change/merge warning. The

forward pedestrian impact avoidance category included pedestrian automatic emergency braking

(PAEB). Lastly, the backing collision avoidance category included rear automatic braking

(RAB) and rear cross-traffic alert (RCTA). These ADAS technologies are characterized as SAE

International (SAE) Level 0-1

driving automation systems.

NHTSA derived target crash populations for each of the five technology categories using

2011 to 2015 Fatality Analysis Reporting System (FARS) and National Automotive Sampling

System General Estimates System (NASS GES) data sets, which serve as records of police-

reported fatal and non-fatal crashes, respectively, on the nation’s roads. For a given technology

category, we compiled data for each of the corresponding pre-crash scenarios to generate target

crash populations surrounding the number of crashes, fatalities, non-fatal injuries, and property-

damage-only vehicles (PDOVs).

See Table 1 for a breakdown of target crash populations for

each technology category.

The study uses the term “lane keeping assist” (LKA), but NCAP terminology differs. NCAP uses the term “lane

keeping support” throughout this document instead.

Similarly, the study uses the term “blind spot detection” (BSD) but NCAP uses the term blind spot warning

(BSW) throughout this document instead.

SAE International (2018), Taxonomy and definitions for terms related to driving automation systems for on-road

motor vehicles (SAE J3016). Level 0: No Automation—The full-time performance by the human driver of all

aspects of the dynamic driving task, even when enhanced by warning or intervention systems. Level 1: Driver

Assistance—The driving mode-specific execution by a driver assistance system of either steering or

acceleration/deceleration using information about the driving environment and with the expectation that the human

driver performs all remaining aspects of the dynamic driving task.

PDOVs are vehicles damaged in non-injury-producing crashes (i.e., crashes in which vehicles only incur property

damage and no occupants incur injury).

Table 1: Summary of Target Crashes by Technology Group

Safety Systems Crashes Fatalities

MAIS 1-5

Injuries

PDOVs

1 FCW/DBS/CIB

1,703,541

29.4%

1,275

3.8%

883,386

31.5%

2,641,884

36.3%

2 LDW/LKA/LCA

1,126,397

19.4%

14,844

44.3%

479,939

17.1%

863,213

11.9%

3 BSW/BSI/LCM

503,070

8.7%

542

1.6%

188,304

6.7%

860,726

11.8%

4 PAEB

111,641

1.9%

4,106

12.3%

104,066

3.7%

6,985

0.1%

5 RAB/RvAB

/RCTA

148,533

2.6%

0.2%

35,268

1.3%

231,317

3.2%

Combined

3,593,182

62%

20,841

62.2%

1,690,963

60.3%

4,604,125

63.3%

It is important to note that target crash populations for the five technology categories

covered 62 percent of all crashes. Crossing path crashes, which also represented a large crash

population and a significant number of fatalities, were not part of our analysis because we are not

aware of a currently available ADAS technology that can effectively mitigate this crash type.

However, there are emerging safety countermeasures that hold potential to address some portion

of these crashes in the future and these technologies will be considered for NCAP as they

mature. These include intersection safety assist (ISA) systems that use onboard sensors with a

wide field of view (e.g. cameras, lidar, radar) as well as vehicle communications systems.

30,31

Defined as reverse automatic braking in DOT HS 812 653.

In its 2019 report, Volpe found that of the 5,480,886 light vehicle crashes occurring from 2011 through 2015,

crossing path crashes, which totaled 1,131,273, represented 21 percent of all light vehicle crashes and 16 percent

(3,972) of all fatalities (25,350).

NHTSA recognizes that ISA systems are currently available on a small number of light vehicles. However,

preliminary NHTSA testing has shown that current-generation ISA systems have only limited capabilities and

therefore would not effectively mitigate intersection-related crashes at this time – which is one of the requirements

in the four prerequisites for inclusion in NCAP.

Vehicle-to-vehicle (V2V) and vehicle-to-everything (V2X) technologies have the potential to address crossing

path crashes, but, while NHTSA remains strongly interested in these technologies, they are not included in the

current roadmap. NHTSA is continuing to consider the various issues that bear upon the deployment path of V2X,

including technological evolution and regulatory changes to the radio spectrum environment.

Loss-of-control in single-vehicle crashes

also had a relatively high target population and

fatality rate,

but were not included because, aside from electronic stability control (ESC)

systems, which are mandated,

the Agency is not aware of an ADAS technology that effectively

prevents this crash type and also meets NHTSA’s criteria for inclusion in NCAP at this time.

Of the pre-crash typologies included in NHTSA’s March 2019 study, rear-end collisions

were found to be the most common crash type with an annual average of 1,703,541 crashes.

Rear-end collisions represented 29.4 percent of all annual crashes (5,799,883), followed by lane

keeping typologies (1,126,397 crashes or 19.4 percent), and those relating to blind spot detection

(503,070 crashes or 8.7 percent). Backing crashes (148,533) represented 2.6 percent of all

crashes, followed by forward pedestrian crashes (111,641) at 1.9 percent.

Rear-end collisions also had the highest number of Maximum Abbreviated Injury Scale

(MAIS)

1-5 injuries at 883,386, which represented 31.5 percent of all non-fatal injuries

(2,806,260) in Table A-1. Lane keeping crashes had the second highest number of injuries at

479,939 (17.1 percent), as shown in Table A-2, and blind spot crashes had the third highest at

188,304 (6.7 percent), as shown in Table A-3. These typologies were followed by forward

pedestrian crashes at 3.7 percent and backing crashes at 1.3 percent, as shown in Table A-4.

37, 38

Crash scenarios were categorized by the first sequence of a crash event. Target crashes for a technology (e.g.,

lane-keeping crashes) were a collective of crash scenarios that are relevant to the technology. The Loss-of-control

in single-vehicle scenario was defined as crashes where the first event was initiated by a passenger vehicle, and the

event was coded as jackknife or traction loss. This crash scenario is mutually exclusive from those included in the

lane-keeping crashes.

Loss-of-control in single-vehicle crashes are about 1% of crashes and associated with 3% of fatalities.

Federal Motor Vehicle Safety Standard No. 126.

In its 2019 report, Volpe categorized 9 percent (470,733) of all light vehicle crashes (5,480,886) occurring from

2011 through 2015 as control loss crashes. Furthermore, 18 percent (4,456) of all fatal crashes (25,350) were due to

control loss.

The Abbreviated Injury Scale (AIS) is a classification system for assessing impact injury severity developed and

published by the Association for the Advancement of Automotive Medicine and is used for coding single injuries,

assessing multiple injuries, or for assessing cumulative effects on more than one injury. AIS ranks individual

injuries by body region on a scale of 1 to 6 where 1=minor, 2=moderate, 3=serious, 4=severe, 5=critical, and

6=maximum (untreatable). MAIS represents the maximum injury severity, or AIS level, recorded for an occupant

(i.e., the highest single AIS for a person with one or more injuries). MAIS 0 means no injury.

The study uses the term “impacts” but for consistency purposes, NCAP uses the term “crashes” in this paragraph.

The Agency notes that the highest number of serious injuries (i.e., MAIS 3 – 5 injuries) were recorded for lane

keeping crashes (21,282 or 0.76 percent of all non-fatal injuries), followed by rear-end crashes (17,918 or 0.64

NHTSA found that the lane keeping technology category, represented by rollover, lane

departure, and road departure crashes, included the highest number of fatalities: 14,844, or 44.3

percent of all fatalities (33,477), as shown in Table A-2. This was followed by the forward

pedestrian impact category, which included 4,106 pedestrian fatalities (12.3 percent), as shown

in Table A-4. The forward collision prevention category, made up of rear-end crashes, included

1,275 fatalities (3.8 percent), as shown in Table A-1.

The blind spot detection technology

category, represented by lane change/merge crashes, accounted for 1.6 percent of all fatalities, as

shown in Table A-3. This was followed by backing crashes at 0.2 percent, as shown in Table A-

5, which defined the backing collision avoidance category. The Agency notes that forward

pedestrian crashes, which comprised the forward pedestrian impact category, ranked second

highest for fatalities, and were the deadliest based on frequency of fatalities per crash.

In selecting the ADAS technologies to include in this proposal, the Agency wanted not

only to target the most frequently occurring crash types, but also prioritize the most fatal and

highest risk crashes. Based on the target crash populations studied, NHTSA believes that those

represented by the forward collision prevention, lane keeping, blind spot detection, and forward

pedestrian impact technology categories account for the most significant safety need.

The Agency notes that ADAS technologies representing the backing collision avoidance

category (i.e., RAB, RvAB, and RCTA) are not being proposed for this program update. The

backing collision avoidance category did not appear in the top third for number of crashes,

number of fatalities, or number of MAIS 1-5 injuries. This may be due, in part, to the fact that a

significant part of this crash target population is addressed by FMVSS No. 111, “Rear

visibility.”

The Agency needs additional time to assess all available real-world data and study

percent), forward pedestrian crashes (5,973 or 0.21 percent), blind spot crashes (3,476 or 0.12 percent), and backing

crashes (454 or 0.02 percent).

Similarly, the study uses the term “impacts” but for consistency purposes, NCAP uses the term “crashes” in this

paragraph.

49 CFR 571.111. See 79 FR 19177 (Apr. 07, 2014).

the effects of the recent full implementation of FMVSS No. 111 prior to considering adoption of

ADAS technologies designed to prevent backing crashes in NCAP. Furthermore, while the

Agency acknowledges that it previously proposed adding rear automatic braking (RAB) to

NCAP in the December 2015 notice, it is continuing to make changes to the RAB test procedure

published in support of that proposal to address the comments received. Thus, it is not proposing

to add this technology to NCAP at this time. The Agency may propose adding to NCAP ADAS

technologies that address the backing pre-crash typologies as the Agency continues to analyze

the real-world data and refine test procedure revisions.

Units of measure contained within this notice include meters (m), kilometers (km),

millimeters per second (mm/s), meters per second (m/s), kilometers per hour (kph), feet (ft.),

inches per second (in./s), feet per second (ft./s), miles per hour (mph), seconds (s), and kilograms

(kg).

A. Lane Keeping Technologies

A study of the 2005 through 2007 fatal crashes

from the National Motor Vehicle Crash

Causation Study (NMVCCS)

identified that 42 percent of lane departure crashes (i.e., where

the driver left the lane of travel prior to the crash) resulted in a rollover and 37 percent resulted in

an opposite direction crash.

After analyzing NHTSA’s 2019 target population study, NHTSA believes that lane

keeping technologies such as lane departure warning (LDW), lane keeping support (LKS), and

lane centering assist (LCA), can address ten pre-crash scenarios including the prevention or

mitigation of roadway departures and crossing the centerline or median (i.e., opposite direction

Wiacek, C., Fikenscher, J., Forkenbrock, G., Mynatt, M., & Smith, P. (2017), Real-world analysis of fatal run-out-

of-lane crashes using the National Motor Vehicle Crash Causation Survey to assess lane keeping technologies, 25

International Conference on the Enhanced Safety of Vehicles, Detroit, Michigan. June 2017, Paper Number 17-

0220.

The National Motor Vehicle Crash Causation Survey (NMVVCS) was a nationwide survey of 5,471 crashes

involving light passenger vehicles, with a focus on factors related to pre-crash events, which were investigated by

the U.S. Department of Transportation and NHTSA over a 2.5-year period from July 3, 2005, to December 31, 2007.

crashes). These pre-crash scenarios represented on average 1.13 million crashes annually or 19.4

percent of all crashes that occurred on U.S. roadways, and resulted in 14,844 fatalities and

479,939 MAIS 1-5 injuries, as shown in Table A-2. This equals 44.3 percent of all fatalities and

17.1 percent of all injuries recorded.

43,44

NCAP currently provides information on the performance of LDW, one of the lane

keeping ADAS technologies. LDW was introduced in the program in 2010 for model year 2011

vehicles.

At the time, the fitment rate for LDW was less than 0.2 percent. In model year 2018,

it was 30.1 percent.

Although the adoption rate for LDW has increased over this period, it has

not increased as significantly as the fitment rate for forward collision warning (FCW), which saw

an approximate 40 percent increase over the same time period. A possible explanation regarding

the lower fitment rate for LDW will be discussed in the next section. A second lane keeping

ADAS technology that the Agency believes is appropriate for inclusion in NCAP is LKS.

NHTSA believes that LKS may provide additional safety benefits that LDW cannot and may

more effectively address the number of fatalities and injuries related to lane departure crashes.

1. Updating Lane Departure Warning (LDW)

Lane departure warning is a NHTSA-recommended technology that is currently included

in NCAP to mitigate lane departure crashes. LDW systems are used to help prevent crashes that

result when a driver unintentionally allows a vehicle to drift out of its lane of travel. These

systems often use camera-based sensors to detect lane markers, such as solid lines (including

those marked for bike lanes), dashed lines, or raised reflective indicators such as Botts’ Dots,

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

When only serious injuries (i.e., MAIS 3 – 5 injuries) were considered, lane keeping crashes represented the

highest number of non-fatal injuries (21,282 or 0.76 percent of all non-fatal injuries), followed by rear-end crashes

(17,918 or 0.64 percent), forward pedestrian crashes (5,973 or 0.21 percent), blind spot crashes (3,476or 0.12

percent), and backing crashes (454or 0.02 percent).

73 FR 40016 (July 11, 2008).

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

ahead of the vehicle.

Lane departure alerts are presented to the driver when the system detects

that the vehicle is laterally approaching or crossing the lane markings. The alert may be visual,

audible, and/or haptic in nature. Visual alerts may show which side of the vehicle is departing

the lane, and haptic alerts may be presented as steering wheel or seat vibrations to alert the

driver. It is expected that an LDW alert will warn the driver of the unintentional lane shift so the

driver can steer the vehicle back into its lane. When a turn signal is activated, the LDW system

acknowledges that the lane change is intentional and does not alert the driver.

As NHTSA continues its assessment of LDW systems under NCAP, it plans to use the

current NCAP test procedure titled, “Lane Departure Warning System Confirmation Test and

Lane Keeping Support Performance Documentation,” dated February 2013.

This protocol

assesses the system’s ability to issue an alert in response to a driving situation intended to

represent an unintended lane departure and to quantify the test vehicle’s position relative to the

lane line at the time of the LDW alert. In NCAP’s LDW tests, a test vehicle is accelerated from

rest to a test speed of 72.4 kph (45 mph) while travelling in a straight line parallel to a single lane

line comprised of one of three marking types: continuous white lines, discontinuous (i.e., dashed)

yellow lines, or discontinuous raised pavement markers (i.e., Botts’ Dots). The test vehicle is

driven such that the centerline of the vehicle is approximately 1.8 m (6 ft.) from the lane edge.

This path must be maintained, and the test speed must be achieved, at least 61.0 m (200 ft.) prior

to the start gate. Once the driver reaches the start gate, he or she manually inputs sufficient

steering to achieve a lane departure with a target lateral velocity of 0.5 m/s (1.6 ft./s) with respect

to the lane line. The driver of the vehicle does not activate the turn signal at any point during the

test and does not apply any sudden inputs to the accelerator pedal, steering wheel, or brake pedal.

Note that performance of LDW systems may be adversely affected by precipitation or poor roadway conditions

due to construction, unmarked intersections, faded/worn/missing lane markings, markings covered with water, etc.

National Highway Traffic Safety Administration. (2013, February). Lane departure warning system confirmation

test and lane keeping support performance documentation. See http://www.regulations.gov, Docket No. NHTSA-

2006-26555-0135.

The test vehicle is driven at constant speed throughout the maneuver. The test ends when the

vehicle crosses at least 0.5 m (1.7 ft.) over the edge of the lane line marking. The scenario is

performed for two different departure directions, left and right, and for all three lane marking

types, resulting in a total of six test conditions. Five repeated trials runs are performed per test

condition.

LDW performance for each test trial is evaluated by examining the proximity of the

vehicle with respect to the edge of a lane line at the time of the LDW alert. The LDW alert must

not occur when the lateral position of the vehicle, represented by a two-dimensional polygon,

greater than 0.8 m (2.5 ft.) from the inboard edge of the lane line (i.e., the line edge closest to the

vehicle when the lane departure maneuver is initiated), and must occur before the lane departure

exceeds 0.3 m (1 ft.). To pass the test, the LDW system must satisfy the pass criteria for three of

the first five valid individual trials

for each combination of departure direction and lane line

type (60 percent) and for 20 of the 30 trials overall (66 percent).

NCAP’s LDW test conditions represent pre-crash scenarios that correspond to a

substantial portion of fatalities and injuries observed in real-world lane departure crashes. In its

independent review of the 2011-2015 FARS and GES data sets, Volpe showed that

approximately 40 and 30 percent of fatalities in fatal road departure and opposite direction

crashes, respectively, occurred when the posted speed was 72.4 kph (45 mph) or less.

Similarly, the data indicated 64 and 63 percent of injuries resulted from road departure and

opposite direction crashes, respectively, that occurred when the posted speed was 72.4 kph (45

mph) or less.

The two-dimensional polygon is defined by the vehicle’s axles in the X-direction (fore-aft), the outer edge of the

vehicle’s tire in the Y-direction (lateral), and the ground in the Z-direction (vertical).

Trial or test trial is a test among a set of tests conducted under the same test conditions (including test speed) with

the same subject vehicle.

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019, August), Statistics of light-

vehicle pre-crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington,

DC: National Highway Traffic Safety Administration.

Although travel speed was unknown or not reported for a high percentage of crashes in

FARS and GES,

when travel speed was reported, approximately 6 and 9 percent of fatal road

departure and opposite direction crashes, respectively, occurred at travel speeds of 72.4 kph (45

mph) or less. Likewise, the data showed 22 and 25 percent of the police-reported non-fatal road

departure and opposite direction crashes, respectively, occurred at 72.4 kph (45 mph) or less.

Volpe’s data review indicates that speeding is prevalent in lane departure relevant pre-crash

scenarios, but most road departure- and opposite direction-related fatalities and injuries did not

occur on highways. For instance, 79 percent of road departure-related fatal crashes and 89

percent of road departure-related police-reported injuries occurred on roads that were not

highways. Similarly, for opposite direction-related crashes, 87 percent of fatalities and 98

percent of injuries did not occur on highways. Because highway driving speeds are on average

much higher than non-highway speeds, the Volpe data about a high percentage of crashes

occurring at speeds under 72.4 kph (45 mph) appears accurate. The test speed of 72.4 kph (45

mph) appears to address a large portion of the travel speeds where the crashes are occurring.

Furthermore, 62 percent of road departure-related fatalities and 76 percent of road

departure-related injuries occurred on straight roads, thereby aligning with NCAP’s test

procedure. For opposite direction-related crashes, 69 percent of fatalities and 67 percent of

police-reported injuries occurred on straight roads.

In its December 2015 notice,

NHTSA expressed concern that the safety benefits

afforded by LDW technology were being diminished due to false activations. Several studies

referenced in that notice had found that drivers were choosing to disable their vehicle’s LDW

system because it was issuing alerts too frequently. The Agency was also concerned about

For road departure crashes, 63 and 68 percent of the travel speed data, respectively, is unknown or not reported in

FARS and GES. For opposite direction crashes, 65 and 67 percent of the data, respectively, is unknown or not

reported in FARS and GES.

80 FR 78522 (Dec. 16, 2015).

missed detections resulting from tar lines reflecting sun light or covered with water and other

unforeseen anomalies that cause unreliable driver warnings. To address these issues and

improve consumer acceptance, NHTSA requested comment in 2015 on whether to revise certain

aspects of NCAP’s LDW test procedure. Specifically, the Agency solicited comment on whether

it is feasible to (1) award NCAP credit to LDW systems that only provide haptic alerts, and (2)

develop additional test scenarios to address false activations and missed detections. The Agency

also proposed to tighten the inboard lane tolerance for its LDW test procedure from 0.8 to 0.3 m

(2.5 to 1.0 ft.). In doing this, an LDW alert could only occur within a window of +0.3 to -0.3 m

(+1.0 to -1.0 ft.) with respect to the inside edge of the lane line to pass NCAP’s LDW procedure.

This proposal effectively increased the space in which a vehicle could operate within a lane

before triggering of an LDW alert was permitted. Each of these topics are discussed in detail in

the sections that follow.

a. Haptic Alerts

With respect to haptic warnings, NHTSA mentioned in its December 2015 notice that

these alerts may offer greater consumer acceptance compared to audible alerts, and thus improve

the effectiveness of LDW alerts if the driver does not view the alerts as a nuisance and disengage

the system. In response to the notice, commenters generally did not support a haptic alert

requirement. Some commenters suggested that requiring a specific feedback type would

unnecessarily limit the manufacturer’s flexibility to issue warnings to the driver, particularly

when considering the potential effectiveness of different feedback types and the need to optimize

human-machine interface (HMI) designs to address a suite of ADAS. Bosch suggested the

Agency should allow all warning options to promote the availability of such systems in a greater

number of vehicles, which should ultimately increase consumer awareness and encourage

vehicle safety improvements. Advocates stated that the Agency should provide details on the

effectiveness of the different types of sensory feedback (visual, auditory, haptic) to justify its

decision to encourage one warning type over another. Consumers Union (CU) suggested

awarding credit for all LDW feedback types and awarding additional points or credit for haptic

alerts to encourage this feedback type in the future. The Automotive Safety Council (ASC)

acknowledged that haptic warnings may improve driver acceptance of LDW systems but

suggested that false activations must also be reduced to realize improved consumer acceptance

and additional safety benefits.

In a large-scale telematics-based study conducted by UMTRI

for NHTSA on LDW

usage, researchers investigated driver behavior in reaction to alerts. Two types of vehicles were

included in the study: vehicles with audible-only alerts and vehicles where the driver had the

option to select either an audible or haptic alert. When the latter was available, the driver

selected the haptic warning 90 percent of the time. Otherwise, the LDW system was turned “off”

38 percent of the time and thus was not providing alerts. For the system that only provided the

audible warning, the LDW was turned “off” 71 percent of the time.

Based on the findings from the UMTRI’s research, NHTSA concludes that haptic alerts

improve driver acceptance of LDW systems. However, the Agency is not certain if an increase

in driver acceptance will translate to an improvement in the overall efficacy of the LDW system

in reducing crashes. Furthermore, NHTSA does not want to hinder optimization of HMI designs

given the increasing number of ADAS technologies available in vehicles today. Therefore, the

Agency has decided not to require a specific alert modality for LDW warnings in its related

NCAP test procedure at this time, but is requesting comment on whether this decision is

appropriate. Although NHTSA has limited data on the effectiveness of the various alert types, it

has some concern (similar to the one raised for FCW) that certain LDW systems, such as those

Flannagan, C., LeBlanc, D., Bogard, S., Nobukawa, K., Narayanaswamy, P., Leslie, A., Kiefer, R., Marchione,

M., Beck, C., and Lobes, K. (2016, February), Large-scale field test of forward collision alert and lane departure

warning systems (Report No. DOT HS 812 247), Washington, DC: National Highway Traffic Safety

Administration.

that may provide only a visual alert, may be less effective than other alert options in medium or

high urgency situations.

b. False Positive Tests

In responding to the 2015 RFC, vehicle manufacturers and suppliers asserted that

additional false positive test requirements were not needed even though they acknowledged

NHTSA’s concern regarding the effect of nuisance alerts on consumer acceptance. Specifically,

the Alliance

stated that vehicle manufacturers will optimize their systems to minimize false

positive activations for consumer acceptance purposes, and thus such tests will not be necessary.

Similarly, Honda stated that vehicle manufacturers must already account for false positives when

considering marketability and HMI. The manufacturer also indicated that it would be difficult

for the Agency to create a valid false positive test procedure that is robust and repeatable.

Mobileye, Bosch, and MTS Systems Corporation (MTS) also agreed. In fact, Mobileye

explained that it would be hard to reproduce the exact test conditions, especially with respect to

weather, over multiple test locations. Also, Bosch stated that the specialized tests required to

address the Agency’s concern may not be truly representative of all real-world driving situations

that the system encounters. MTS suggested that, alternatively, a new test could be added to

NCAP’s LDW test procedure that would evaluate whether an LDW system can inform the driver

that it is no longer able to issue warnings due to poor environmental conditions or other reasons.

Given the concerns expressed regarding repeatability and reproducibility of test

conditions, and the fact that the Agency’s data do not currently support adoption of a false

Lerner, N., Robinson, E., Singer, J., Jenness, J., Huey, R., Baldwin, C., & Fitch, G. (2014, September), Human

factors for connected vehicles: Effective warning interface research findings (Report No. DOT HS 812 068),

Washington, DC: National Highway Traffic Safety Administration.

After submitting individual comments on the 2015 RFC, the Alliance and Global Automakers merged to form the

Alliance for Automotive Innovation. This document addresses the individual comments from the organizations that

were then the Alliance and Global Automakers.

positive assessment for lane keeping technologies, NHTSA continues to monitor the consumer

complaint data related to false positives to help inform an appropriate next step.

With respect to the recommendation from MTS, the Agency recognizes that vehicle

manufacturers install LDW telltales on the instrument panel that illuminate to inform drivers

when the system is operational. The systems are typically operational when the vehicle’s travel

speed has reached a preset activation threshold speed and the lane markings and environmental

conditions are appropriate. The telltale will disappear if those conditions are not met to inform

the driver that the system is no longer operational. In such a state, the system will not provide an

alert if the vehicle departs the travel lane. Given this feature, NHTSA has decided a test to

inform the driver that the system is no longer issuing warnings is unnecessary at this time.

c. LDW Test Procedure Modifications

Support was varied with respect to NHTSA’s proposal in the December 2015 notice to

modify the LDW test requirements to reduce the leeway for system activation inside of a lane

line from 0.8 to 0.3 m (2.5 to 1.0 ft.). Global Automakers stated that the proposed change was

“unduly prescriptive” and recommended that the Agency retain the existing lane line tolerance.

The organization explained that research showed 90 percent of drivers needed 1.2 s to react to a

warning.

Citing NCAP’s LDW test procedure, which requires a steering input having a target

lateral velocity of 0.5 to 0.6 m/s (1.6 to 2 ft./s), the trade association remarked that this

requirement equates to a necessary warning distance of 0.6 to 0.72 m (1.9 to 2.4 ft.) to ensure

that 90 percent of drivers can react in time to prevent a lane departure. Advocates agreed that

nuisance notifications are a concern for driver acceptance, but noted that the Agency provided

little information about the effectiveness of LDW systems meeting the proposed criteria.

Conversely, Delphi, ASC, and MTS commented that some of the more robust systems that are

Tanaka, S., Mochida, T., Aga, M., & Tajima, J. (2012, April 16). Benefit Estimation of a Lane Departure Warning

System using ASSTREET. SAE Int. J. Passeng. Cars - Electron. Electr. Syst. 5(1):133-145, 2012,

https://doi.org/10.4271/2012-01-0289.

currently available should be able to comply with the narrower specification. However, ASC

suggested that the Agency may want to evaluate the impact of the proposed changes before

finalizing the requirements to ensure that narrowing the lane line tolerances translates to a

reduction in false positive alerts, and thus higher consumer acceptance for LDW systems.

Mobileye stated that the tolerance reduction should increase the required accuracy and quality of

lane keeping systems. MTS remarked that systems meeting the tighter specification will produce

higher driver satisfaction, and, in turn, system use, compared to those that meet only the current

requirements. Hyundai Motor Company (Hyundai) also supported the tolerance revision.

Consumers Union (CU) agreed with others that the narrowed lateral tolerance should reduce the

issuance of false alerts on main roadways but cautioned the Agency that this change may not

effectively address false alerts on secondary or curved roads, as vehicles not only tend to

approach within one foot of lane lines, but also may cross them. The group suggested that false

alert conditions be subject to speed limitations or GPS-based position sensors to avoid “over

activation” on secondary or curved roads.

Given NHTSA’s goal of reducing nuisance notifications to increase consumer acceptance

of LDW systems and the statements from several commenters that current LDW systems can

meet the proposed reduced test specification, the Agency believes it is reasonable to propose

adopting the reduced inboard lane tolerance of 0.3 m (1.0 ft.).

In addition to the comments received pertaining to the lane line tolerance, the Agency

also received several suggestions to adopt additional test scenarios for NCAP’s LDW test

procedure or make alternative procedural modifications. Similar to CU’s suggestion above for

curved roads, Mobileye suggested that NHTSA add inner and outer curve scenarios that allow a

larger tolerance for the inner lane boundary than that permitted on a straight road. The company

further recommended that the Agency add road edge detection scenarios, including curbs and

non-structural delimiters such as gravel or dirt, to reflect real-world conditions and crash

scenarios more accurately. Similarly, Bosch suggested that NHTSA consider introducing road

edge detection requirements in addition to lane markings since not all roads have lane markings.

Additionally, Mobileye suggested that NHTSA alter the Botts’ Dots detail #4 (Botts dots are

round, raised markers that mark lanes) to align with California detail #13, which is more

common, and modify the test procedure to include Botts’ Dots on both sides of the lane or Botts’

Dots and a solid line, as these are the most frequently observed marking pairings.

The Agency appreciates suggestions from commenters and agrees that there is merit to

considering other procedural modifications for NCAP’s lane departure test procedure(s). As will

be discussed in the next section, the Agency is planning to conduct a feasibility study to

determine whether curved roads can be considered for inclusion in NCAP test procedures to

evaluate LKS systems objectively. NHTSA also plans to perform research to assess how lane

keeping system performance on a test track compares to real-world data for different

combinations of curve radius, vehicle speed, and departure timing. Additionally, the Agency

recognizes that the European NCAP program (Euro NCAP) has adopted a road edge detection

test that is conducted in a similar manner to their “lane keep assist” tests (described in the next

section), but the road edge detection test does not use lane markings. Although NHTSA believes

the number of vehicles equipped with an ability to recognize and respond to road edges not

defined with a lane line is presently low, it has identified roadways where this capability could

prevent crashes. Therefore, the Agency is requesting comment on whether a road edge detection

test for either LDW and/or LKS is appropriate for inclusion in NCAP. In consideration of the

lane markings currently assessed, the Agency proposes to remove the Botts’ Dots test scenario

from the current LDW test, as the lane marking type is being removed from use in California.

Winslow, J. (2017, May 19), Botts’ Dots, after a half-century, will disappear from freeways, highways, The

Orange County Register, https://www.ocregister.com/2017/05/19/botts-dots-after-a-half-century-will-disappear-

from-freeways-highways/.

At this time, the Agency believes the traditional dashed and solid lane marking tests would be

sufficient.

Although NHTSA has tentatively decided not to adopt additional false activation

requirements for this NCAP upgrade, the Agency is still concerned about the low effectiveness

of LDW and its lack of consumer acceptance stemming from nuisance alerts and missed

detections.

When NHTSA decided to include ADAS in the NCAP program in 2008,

LDW was

selected because it met NCAP’s four established criteria: (1) the technology addressed a major

crash problem; (2) the system design of LDW had the potential to mitigate the crash problem; (3)

safety benefits were projected, and (4) test procedures and evaluation criteria were available to

ensure an acceptable performance level. At the time, the Agency estimated that existing LDW

systems were 6 to 11 percent effective in preventing lane departure crashes. Although the

system’s effectiveness was relatively low, NHTSA cited the large number of road departure and

opposite direction crashes occurring on the nation’s roadways as well as the resulting AIS 3+

injuries, as reasons to include LDW in NCAP. Several recent studies have provided varying

results with respect to LDW effectiveness.

In a 2017 study,

the Insurance Institute for Highway Safety (IIHS) concluded that LDW

systems were effective in reducing three types of passenger car crashes (single-vehicle, side-

swipes, and head-on) by 11 percent, which is the same rate NHTSA originally estimated.

Importantly, IIHS also concluded that LDW systems reduce injuries in those same types of

crashes by 21 percent. In its recent study of real-world effectiveness of crash avoidance

73 FR 40033 (July 11, 2008).

Insurance Institute for Highway Safety (2017, August 23), Lane departure warning, blind spot detection help

drivers avoid trouble, https://www.iihs.org/news/detail/stay-within-the-lines-lane-departure-warning-blind-spot-

detection-help-drivers-avoid-trouble.

technologies in GM vehicles,

UMTRI found that LDW systems showed a 3 percent reduction

for applicable crashes that was determined to be not statistically significant. Conversely, the

active safety technology, LKS (which also included lane departure warning capability), showed

an estimated 30 percent reduction in applicable crashes.

Other studies that examined driver deactivation rates also suggest that LDW

effectiveness may be lower than originally estimated. In a survey of Honda vehicles brought into

Honda dealerships for service,

IIHS researchers found that for 184 models equipped with an

LDW system, only a third of the vehicles had the system activated. Furthermore, in its

telematics-based study on LDW usage,

UMTRI found that, overall, drivers turned off LDW

systems 50 percent of the time. However, in Consumer Reports’ August 2019 survey of more

than 57,000 CR subscribers, the organization found that 73 percent of vehicle owners reported

that they were satisfied with LDW technology. In fact, 33 percent said that the system had

helped them avoid a crash, and 65 percent said that they trusted the system to work every time.

In light of these findings, the Agency believes that, in addition to LDW, there is merit to

adopting an active lane keeping system, such as lane keeping support (LKS), in NCAP. As an

enhanced active system, LKS offers the steering and/or braking capability necessary to guide a

vehicle back into its lane without consumer action and should therefore further enhance safety

benefits beyond those that can be realized by LDW. A detailed discussion pertaining to LKS

technology is provided in the following section.

Flannagan, C. and Leslie, A., Crash Avoidance Technology Evaluation Using Real-World Crashes,

DTHN2216R00075 Vehicle Electronics Systems Safety IDIQ, The University of Michigan Transportation Research

Institute Final Report, March 22, 2018.

Insurance Institute for Highway Safety (2016, January 28), Most Honda owners turn off lane departure warning,

Status Report, Vol. 51, No. 1, page 6.

Flannagan, C., LeBlanc, D., Bogard, S., Nobukawa, K., Narayanaswamy, P., Leslie, A., Kiefer, R., Marchione,

M., Beck, C., and Lobes, K. (2016, February), Large-scale field test of forward collision alert and lane departure

warning systems (Report No. DOT HS 812 247), Washington, DC: National Highway Traffic Safety

Administration.

Consumer Reports (2019, August 5), Guide to lane departure warning & lane keeping assist: Explaining how

these systems can keep drivers on the right track, https://www.consumerreports.org/car-safety/lane-departure-

warning-lane-keeping-assist-guide/.

2. Adding Lane Keeping Support (LKS)

LDW systems warn a driver that their vehicle is unintentionally drifting out of their travel

lane, while lane keeping support (LKS) systems are designed to actively guide a drifting vehicle

back into the travel lane by gently counter steering or applying differential braking. During an

unintended lane departure where the driver is not using the turn signal, LKS systems help to

prevent: “sideswiping” where a vehicle strikes another vehicle in an adjacent lane that is

travelling in the same direction; opposite direction crashes where a vehicle crosses the centerline

and strikes another vehicle travelling in the opposite direction; and road departure crashes where

a vehicle runs off the road resulting in a rollover crash or an impact with a tree or other object.

LKS systems may also help to prevent unintended lane departures into designated bicycle lanes

in situations where the system’s speed threshold is met. 

LKS systems typically utilize the same camera(s) used by LDW systems to monitor the

vehicle’s position within the lane, and determine whether a vehicle is about to drift out of its lane

of travel unintentionally. In such instances, LKS automatically intervenes by: braking one or

more of the vehicle’s wheels; steering; or using a combination of braking and steering so that the

vehicle returns to its intended lane of travel. LKS is one of two active lane keeping technologies

mentioned in the Agency’s March 2019 report,

with the other being lane centering assist

(LCA). LKS assists the driver by providing short-duration steering and/or braking inputs when a

lane departure is imminent or underway, whereas LCA provides continuous assistance to the

driver to keep their vehicle centered within the lane.

As discussed in the previous section, UMTRI evaluated the real-world effectiveness of

ADAS technologies, including LDW and LKS.

The results of the LKS study (which also

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

Carol Flannagan, Andrew Leslie, Crash Avoidance Technology Evaluation Using Real-World Crashes,

DTHN2216R00075 Vehicle Electronics Systems Safety IDIQ, The University of Michigan Transportation Research

Institute Final Report, March 22, 2018.

included lane departure warning functionality) showed an estimated 30 percent reduction in

applicable crashes. Additionally, in its August 2019 survey, 74 percent of vehicle owners

reported that they were satisfied with LKS technology, and 35 percent said that it had helped

them avoid a crash. Sixty-five percent of owners said that they trusted the system to work every

time.

In its December 2015 notice, NHTSA did not propose including LKS technology as part

of the update to NCAP. However, many commenters recommended that the Agency consider

including the technology. For instance, Bosch and Mobileye stated that LKS systems have the

potential to prevent or mitigate a greater number of collisions involving injuries and fatalities

than LDW systems. The ASC and Delphi recommended that the Agency adopt LKS in lieu of

LDW, with the ASC adding that Euro NCAP has included LKS in its Lane Support Systems test

protocol since 2016.

68,69

The ASC, Bosch, and Continental noted the maturity of LKS

technology and stated that such systems were already widely available in vehicles produced at

the time. Other proponents of adopting LKS technology in NCAP include the National Safety

Council (NSC), ZF TRW, and Honda. ZF TRW recommended that the Agency adopt both

active lane keeping (termed LKS in this notice) and lane centering systems (termed LCA in this

notice) due to the high frequency of fatal road departure crashes. Honda also supports the active

safety benefits of LKS and the system’s potential to help prevent crashes. NSC suggested that

the Agency include LKS, as it would complement LDW, which is already in the program, similar

to the way the warning component of FCW complements the active safety functionality of AEB.

Consumer Reports. (2019, August 5), Guide to lane departure warning & lane keeping assist: Explaining how

these systems can keep drivers on the right track, https://www.consumerreports.org/car-safety/lane-departure-

warning-lane-keeping-assist-guide/.

The ASC argued that data from the Highway Loss Data Institute (HLDI) have shown no statistically significant

difference in collision claim frequencies for vehicles equipped with LDW compared to those without, and

questioned whether LDW systems are effective in reducing crashes or fatalities.

European New Car Assessment Programme (Euro NCAP) (2015, November), Test Protocol – Lane Support

Systems, Version 1.0.

As mentioned previously, the Agency agrees with commenters that there is merit to

adopting LKS technology in NCAP. However, NHTSA believes an LDW system integrated

with LKS may be a better approach for the Agency to consider rather than replacing LDW with

LKS. NHTSA believes, as NSC commented, that an integrated approach (inclusive of passive

and active safety capabilities for lane support systems) would be similar to what the Agency is

proposing for frontal collision avoidance systems, FCW and AEB, later in this notice.

NHTSA is considering the adoption of certain test methods (e.g., those for “lane keep

assist”) contained within the Euro NCAP Test Protocol - Lane Support Systems (LSS)

to assess

technology design differences for LKS. Since the test speeds and road configurations specified

in this protocol are similar to those stipulated in the Agency’s LDW test procedure, the Agency

believes Euro NCAP’s test protocol will sufficiently address the lane keeping crash typology

previously detailed for LDW.

Euro NCAP’s LSS test procedure includes a series of “lane keep assist” trials that are

performed with iteratively increasing lateral velocities towards the desired lane line. Each “lane

keep assist” trial begins with the subject vehicle (SV) (i.e., the vehicle being evaluated) being

driven at 72 kph (44.7 mph) down a straight lane delineated by a single solid white or dashed

white line. Initially, the SV path is parallel to the lane line, with an offset from the lane line that

depends on the lateral velocity used later in the maneuver. Then, after a short period of steady-

state driving, the direction of travel of the SV is headed towards the lane line using a path

defined by a 1,200 m (3,937.0 ft.) radius curve. The lateral velocity of the SV’s approach

towards the lane line (from both the left and right directions) is increased from 0.2 to 0.5 m/s (0.7

to 1.6 ft./s) in 0.1 m/s (0.3 ft./s) increments until acceptable LKS performance is no longer

European New Car Assessment Programme (Euro NCAP) (2019, July), Test Protocol – Lane Support Systems,

Version 3.0.2. See section 7.2.5, Lane Keep Assist tests.

realized. Acceptable LKS performance occurs when the SV does not cross the inboard leading

edge of the lane line by more than 0.3 m (1.0 ft.).

NHTSA conducted a limited assessment of five model year 2017 vehicles equipped with

LKS systems. The Agency used a robotic steering controller to maximize the repeatability and

minimize variability associated with manual steering inputs. For this study, NHTSA also used a

slightly modified and older version of Euro NCAP’s LSS test procedure from what was

discussed above. Specifically, the lateral velocity of the SV’s approach towards the lane line

was increased from 0.1 m/s to 1.0 m/s in 0.1 m/s increments (0.3 ft./s to 3.3 ft./s in 0.3 ft./s

increments) to assess how LKS systems would perform at higher velocities. In addition, LKS

performance was considered acceptable (when compared to Euro NCAP’s assessment criteria at

the time of NHTSA’s testing) for instances where the SV did not cross the inboard leading edge

of the lane line by more than 0.4 m (1.3 ft.).

A preliminary analysis of the five tested vehicles identified performance differences

between the vehicles depending on the lateral velocity used during the test. Some vehicles only

engaged a steering response at lower lateral velocities and others continued to provide a steering

input as the lateral velocity was increased.

The maximum excursion over the lane marking

after an LKS activation was also found to be inconsistent, particularly as lateral velocity

increased. These preliminary findings suggested that there are performance differences in how

vehicle manufacturers are designing their systems for a given set of operating conditions.

The results from these tests, as measured by the maximum excursions over the lane

marking, were compared to the measured shoulder width of roads where fatal road departure

At the time of testing, an older version of Euro NCAP’s LSS test procedure was available. This version stipulated

a lane keep assist assessment criterion of 0.4 m (1.3 ft.) for the maximum excursion over the inside edge of the lane

marking. European New Car Assessment Programme (Euro NCAP). See Assessment Protocol – Safety Assist,

Version 7.0 (2015, November).

Wiacek, C., Forkenbrock, G., Mynatt, M., & Shain, K. (2019), Applying lane keeping support test track

performance to real-world crash data, 26th Enhanced Safety of Vehicles Conference, Eindhoven, Netherlands. June

2019, Paper Number 19-0208.

crashes occurred. The analysis identified roadways where the shoulder width of the roadway

was less than the 0.4 m (1.3 ft.) maximum excursion limit (e.g., certain rural roadways) used in

the Agency’s testing. It was observed that only vehicles displaying robust LKS performance,

including at higher lateral velocities, would likely prevent the vehicle from departing the travel

lane on these roadways. However, most of the roadway departure crashes were on roads where

the shoulder width exceeded 0.4 m (1.3 ft.). On these roadways, assuming the LKS was

engaged, the lane departure could have been avoided. However, some vehicles did not perform

well, with several exhibiting no system intervention, and others exceeding the maximum

excursion limit as the lateral velocity was increased. To supplement these initial findings,

additional LKS testing has since been conducted and is undergoing analysis.

Since the analysis showed that most fatal crashes identified in the study were on

roadways having shoulder widths that exceeded the current Euro NCAP test excursion limit of

0.3 m (1.0 ft.), NHTSA believes that adopting the Euro NCAP criterion may provide significant

safety benefits, but is requesting comment on whether an even smaller excursion limit may be

more appropriate. Furthermore, as the study also identified fatal crashes where lane markers

were not present on the side of the roadway where a departure occurred (such that LKS would

not provide any benefit unless it had the capability to identify the edge of the roadway), the

Agency is also requesting comment (as mentioned previously) on adding Euro NCAP’s road

edge detection test to NCAP so that it may begin to address crashes that occur where lane

markings may not be present.

Based on the findings from NHTSA’s LKS testing, which showed differences in LKS

performance at greater lateral velocities, the Agency is concerned about LKS performance at

higher travel speeds when the vehicle first transitions from a straight to a curved road where

lateral velocity may inherently be high. In its independent analysis of the 2011-2015 FARS data

set, Volpe found that 29 percent of fatal road departure crashes and 26 percent of fatal opposite

direction crashes occurred at known travel speeds exceeding 72.4 kph (45 mph). The analysis

also showed that 55 percent of fatal road departure crashes and 67 percent of opposite direction

crashes occurred on roads with posted speeds exceeding 72.4 kph (45 mph).

73,74

Furthermore,

the study revealed that speeding was a factor in 31 percent and 13 percent of fatal road departure

and opposite direction crashes, respectively.

Since NHTSA does not currently have data to

show that LKS system performance at Euro NCAP’s current test speed of 72 kph (44.7 mph)

would be indicative of system performance when tested at higher speeds, NHTSA is requesting

comment on whether it would be beneficial to incorporate additional, higher test speeds to assess

the performance of lane keeping systems in NCAP.

To date, NHTSA has only performed test track LKS evaluations using the straight road

test configuration specified in the Euro NCAP test procedure. However, the Agency recognizes

that a significant portion of road departure and opposite direction crashes resulting in fatalities

and injuries occur on curved roads. A review of Volpe’s 2011-2015 data set

showed that for

road departure crashes, 37 percent of fatalities and 20 percent of injuries occurred on curved

roads. For opposite direction crashes, 30 percent of fatalities and 31 percent of injuries occurred

on curved roads. NHTSA is not certain how LKS performance observed during straight road

trials performed on a test track would correlate to real-world system performance on curved

roads. However, NHTSA believes, based on on-road performance testing experience of newer

model year vehicles, that some current system designs include provisions to address lane

departures on curved roads. The Agency observed that some LKS systems engage by providing

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019, August), Statistics of light-

vehicle pre-crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington,

DC: National Highway Traffic Safety Administration.

For data where the travel speed was known, 63 and 65 percent of the data is unknown or not reported in FARS for

road departure and opposite direction crashes, respectively. For road departure and opposite direction crashes,

respectively, 3 and 1 percent of the posted speed data is unknown or not reported in FARS.

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019, August), Statistics of light-

vehicle pre-crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington,

DC: National Highway Traffic Safety Administration.

Ibid.

limited operation throughout a curve—which may offer little (if any) safety benefits. However,

other more sophisticated LKS systems maintain engagement longer and offer more directional

authority throughout a curve. These systems may provide additional safety gains because the

driver has more time to re-engage (i.e., restore effective manual control of the vehicle).

In NHTSA’s study of the 2005 through 2007 fatal crashes

from NMVCCS, crashes that

occurred on curved roads

where the driver departed the travel lane were analyzed. The analysis

showed that, unlike for straight roads where LKS systems may provide smaller corrective

steering inputs to prevent the vehicle from departing the lane, LKS systems would have to

provide sustained lateral correction (i.e., corrective steering) on a curved road to prevent the

vehicle from departing the lane.

Furthermore, in fleet testing of select model year 2012 through 2018 vehicles equipped

with LDW and LKS (referenced in the report as LKA), Transport Canada

found variability in

test results and generally unpredictable system behavior on curved roads. Thus, Transport

Canada stated that it was not possible to gather enough data to assess the potential safety benefits

associated with the technology.

To address these unknowns and further understand the potential effectiveness of LKS

systems in the real world, the Agency is considering additional research to study whether testing

on curved roads should be considered for objective evaluation of LKS systems, and collect a

combination of test track and real-world data to quantify how LKS systems will operate when

Wiacek, C., Fikenscher, J., Forkenbrock, G., Mynatt, M., & Smith, P. (2017), Real-world analysis of fatal run-out-

of-lane crashes using the National Motor Vehicle Crash Causation Survey to assess lane keeping technologies, 25th

International Technical Conference on the Enhanced Safety of Vehicles, Detroit, Michigan. June 2017, Paper

Number 17-0220.

It should be noted that the paper identified crashes where lane markings were not present on the side of the

departure.

Meloche, E., Charlebois, D., Anctil, B., Pierre, G., & Saleh, A. (2019), ADAS testing in Canada: Could partial

automation make our roads safer? 26th International Technical Conference on the

Enhanced Safety of Vehicles,

Eindhoven, Netherlands, June 2019, Paper Number 19-0339.

exposed to different combinations of curve radius, vehicle speed, and departure timing (e.g., at

curve onset or midway through the curve).

With respect to LDW and LKS, NHTSA is seeking comment on the following:

(1) Should the Agency award credit to vehicles equipped with LDW systems that provide a

passing alert, regardless of the alert type? Why or why not? Are there any LDW alert

modalities, such as visual-only warnings, that the Agency should not consider

acceptable when determining whether a vehicle meets NCAP’s performance test

criteria? If so, why? Should the Agency consider only certain alert modalities (such as

haptic warnings) because they are more effective at re-engaging the driver and/or have

higher consumer acceptance? If so, which one(s) and why?

(2) If NHTSA were to adopt the lane keeping assist test methods from the Euro NCAP LSS

protocol for the Agency’s LKS test procedure, should the LDW test procedure be

removed from its NCAP program entirely and an LDW requirement be integrated into

the LKS test procedure instead? Why or why not? For systems that have both LDW

and LKS capabilities, the Agency would simply turn off LKS to conduct the LDW test

if both systems are to be assessed separately. What tolerances would be appropriate for

each test, and why?

(3) LKS system designs provide steering and/or braking to address lane departures (e.g.,

when a driver is distracted).

To help re-engage a driver, should the Agency specify

that an LDW alert must be provided when the LKS is activated? Why or why not?

Cicchino, J. B. & Zuby, D. S. (2016, October), Prevalence of driver physical factors leading to unintentional lane

departure crashes, Traffic Injury Prevention, 18(5), 481-487, https://doi.org/10.1080/15389588.2016.1247446.

(4) Do commenters agree that the Agency should remove the Botts’ Dots test scenario from

the current LDW test procedure since this lane marking type is being removed from use

in California?

If not, why?

(5) Is the Euro NCAP maximum excursion limit of 0.3 m (1.0 ft.) over the lane marking (as

defined with respect to the inside edge of the lane line) for LKS technology acceptable,

or should the limit be reduced to account for crashes occurring on roads with limited

shoulder width? If the tolerance should be reduced, what tolerance would be

appropriate and why? Should this tolerance be adopted for LDW in addition to LKS?

Why or why not?

(6) In its LSS Protocol, Euro NCAP specifies use of a 1,200 m (3,937.0 ft.) curve and a

series of increasing lateral offsets to establish the desired lateral velocity of the SV

towards the lane line it must respond to. Preliminary NHTSA tests have indicated that

use of a 200 m (656.2 ft.) curve radius provides a clearer indication of when an LKS

intervention occurs when compared to the baseline tests performed without LKS, a

process specified by the Euro NCAP LSS protocol. This is because the small curve

radius allows the desired SV lateral velocity to be more quickly established; requires

less initial lateral offset within the travel lane; and allows for a longer period of steady

state lateral velocity to be realized before an LKS intervention occurs. Is use of a 200

m (656.2 ft.) curve radius, rather than 1,200 m (3,937.0 ft.), acceptable for inclusion in

a NHTSA LKS test procedure? Why or why not?

(7) Euro NCAP’s LSS protocol specifies a single line lane to evaluate system performance.

However, since certain LKS systems may require two lane lines before they can be

Winslow, J. (2017, May 19), Botts’ Dots, after a half-century, will disappear from freeways, highways, The

Orange County Register, https://www.ocregister.com/2017/05/19/botts-dots-after-a-half-century-will-disappear-

from-freeways-highways/.

enabled, should the Agency use a single line or two lines lane in its test procedure?

Why?

(8) Should NHTSA consider adding Euro NCAP’s road edge detection test to its NCAP

program to begin addressing crashes where lane markings may not be present? If not,

why? If so, should the test be added for LDW, LKS, or both technologies?

(9) The LKS and “Road Edge” recovery tests defined in the Euro NCAP LSS protocol

specify that a range of lateral velocities from 0.2 to 0.5 m/s (0.7 to 1.6 ft./s) be used to

assess system performance, and that this range is representative of the lateral velocities

associated with unintended lane departures (i.e., not an intended lane change).

However, in the same protocol, Euro NCAP also specifies a range of lateral velocities

from 0.3 to 0.6 m/s (1.0 to 2.0 ft./s) be used to represent unintended lane departures

during “Emergency Lane Keeping – Oncoming vehicle” and “Emergency Lane

Keeping – Overtaking vehicle” tests. To encourage the most robust LKS system

performance, should NHTSA consider a combination of the two Euro NCAP

unintended departure ranges, lateral velocities from 0.2 to 0.6 m/s (0.7 to 2.0 ft./s), for

inclusion in the Agency’s LKS evaluation? Why or why not?

(10) As discussed above, the Agency is concerned about LKS performance on roads that are

curved. As such, can the Agency correlate better LKS system performance at higher

lateral velocities on straight roads with better curved road performance? Why or why

not? Furthermore, can the Agency assume that a vehicle that does not exceed the

maximum excursion limits at higher lateral velocities on straight roads will have

superior curved road performance compared to a vehicle that only meets the excursion

limits at lower lateral velocities on straight roads? Why or why not? And lastly, can

the Agency assume the steering intervention while the vehicle is negotiating a curve is

sustained long enough for a driver to re-engage? If not, why?

(11) The Agency would like to be assured that when a vehicle is redirected after an LKS

system intervenes to prevent a lane departure when tested on one side, if it approaches

the lane marker on the side not tested, the LKS will again engage to prevent a

secondary lane departure by not exceeding the same maximum excursion limit

established for the first side. To prevent potential secondary lane departures, should the

Agency consider modifying the Euro NCAP “lane keep assist” evaluation criteria to be

consistent with language developed for NHTSA’s BSI test procedure to prevent this

issue? Why or why not? NHTSA’s test procedure states the SV BSI intervention shall

not cause the SV to travel 0.3 m (1 ft.) or more beyond the inboard edge of the lane line

separating the SV travel lane from the lane adjacent and to the right of it within the

validity period. To assess whether this occurs, a second lane line is required (only one

line is specified in the Euro NCAP LSS protocol for LKS testing). Does the

introduction of a second lane line have the potential to confound LKS testing? Why or

why not?

(12) Since most fatal road departure and opposite direction crashes occur at higher posted

and known travel speeds, should the LKS test speed be increased, or does the current

test speed adequately indicate performance at higher speeds, especially on straight

roads? Why or why not?

(13) The Agency recognizes that the LKS test procedure currently contains many test

conditions (i.e., line type and departure direction). Is it necessary for the Agency to

perform all test conditions to address the safety problem adequately, or could NCAP

test only certain conditions to minimize test burden? For instance, should the Agency

consider incorporating the test conditions for only one departure direction if the vehicle

manufacturer provides test data to assure comparable system performance for the other

direction? Or, should the Agency consider adopting only the most challenging test

conditions? If so, which conditions are most appropriate? For instance, do the dashed

line test conditions provide a greater challenge to vehicles than the solid line test

conditions?

(14) What is the appropriate number of test trials to adopt for each LKS test condition, and

why? Also, what is an appropriate pass rate for the LKS tests, and why?

(15) Are there any aspects of NCAP’s current LDW or proposed LKS test procedure that

need further refinement or clarification? Is so, what additional refinements or

clarifications are necessary?

B. Blind Spot Detection Technologies

NHTSA’s 2019 target population study showed that blind spot detection technologies

such as blind spot warning (BSW), blind spot intervention (BSI), and lane change/merge

warning (LCM) (which is essentially a BSI warning system), can help prevent or mitigate five

pre-crash lane change/merge scenarios. These pre-crash movements represented, on average,

503,070 crashes annually, or 8.7 percent of all crashes that occurred on U.S. roadways, and

resulted in 542 fatalities and 188,304 MAIS 1-5 injuries, as shown in Table A-3. This equated to

1.6 percent of all fatalities and 6.7 percent of all injuries recorded.

Currently, NCAP does not include any ADAS technology that is designed to address

blind spot pre-crash scenarios. NHTSA requested comment on the inclusion of BSW as part of

its upgrade to the program in its 2015 notice. Although the Agency did not recommend BSI for

inclusion at that time, the Agency is proposing that both BSW and BSI technologies be adopted

as part of this program update.

Although the target population for blind spot detection technology may not be as large as

the populations for AEB or lane keeping technologies, NHTSA believes there is merit to

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

including blind spot technologies in NCAP. Consumer Reports found in its 2019 survey that 82

percent of vehicle owners were satisfied with BSW technology, 60 percent said that it had helped

them avoid a crash, and 68 percent stated that they trusted the system to work every time.

The

Agency believes the technology’s high consumer acceptance rate, in addition to its potential

safety benefits discussed later in this section, supports its inclusion in the Agency’s signature

consumer information program.

1. Adding Blind Spot Warning (BSW)

A BSW system is a warning-based driver assistance system designed to help the driver

recognize that another vehicle is approaching, or being operated within, the blind spot of their

vehicle in an adjacent lane. In these driving situations, and for all production BSW systems

known to NHTSA, the BSW alert is automatically presented to the driver, and is most relevant to

a driver who is contemplating, or who has just initiated, a lane change. Depending on the system

design, additional BSW features may be activated if the system is presenting an alert and then

the driver operates their turn signal indicator.

BSW systems use camera-, radar-, or ultrasonic-based sensors, or some combination

thereof, as their means of detection. These sensors are typically located on the sides and/or rear

of a vehicle. BSW alerts may be auditory, visual (most common), or haptic. Visual alerts are

usually presented in the side outboard mirror glass, inside edge of the mirror housing, or at the

base of the front a-pillars inside the vehicle. When another vehicle enters, or approaches, the

driver’s blind spot while operating in an adjacent lane, the BSW visual alert will typically be

continuously illuminated. However, if the driver engages the turn signal in the direction of the

adjacent vehicle while the visual alert is present, the visual alert may transition to a flashing state

Monticello, M. (2017, June 29), The positive impact of advanced safety systems for cars: The latest car-safety

technologies have the potential to significantly reduce crashes, Consumer Reports,

https://www.consumerreports.org/car-safety/positive-impact-of-advanced-safety-systems-for-cars/.

and/or be supplemented with an additional auditory or haptic alert (e.g., beeping or vibration of

the steering wheel or seat, respectively).

NHTSA requested comment on a draft research blind spot detection (BSD) test procedure

(referred to in this notice as BSW) published on November 21, 2019

to assess systems’

performance and capabilities in blind spot related pre-crash scenarios. This test procedure

exercises the BSW system in two different scenarios on the test track: the Straight Lane

Converge and Diverge Test, and the Straight Lane Pass-by Test. These two tests assess whether

the BSW system displays a warning when other vehicles, referred to as principal other vehicles

(POVs), are within the driver’s blind spot. The test occurs without activation of the tested

vehicle’s, referred to as the subject vehicle (SV), turn signal. Neither the SV nor POV turn

signals are to be activated at any point during any test trial. A short description of each test

scenario and the requirements for a passing result is provided below:

 Straight Lane Converge and Diverge Test – The POV and SV are driven parallel to each

other at a constant speed of 72.4 kph (45 mph) such that the front-most part of the POV is

1.0 m (3.3 ft.) ahead of the rear-most part of the SV in the outbound lanes of a three-lane

straight road. After 2.5 s of steady-state driving, the POV enters (i.e., converges into) the

SV’s blind zone

by making a single lane change into the lane immediately adjacent to

the SV using a lateral velocity of 0.25 to 0.75 m/s (0.8 to 2.5 ft./s). The period of steady-

state driving resumes for at least another 2.5 s and then the POV exits (i.e., diverges

from) the SV’s blind zone by returning to its original travel lane using a lateral velocity

84 FR 64405 (Nov. 21, 2019).

SV blind zones are defined by two rectangular regions that extend to the side and rear of the SV. Each rectangle is

8.2 ft. (2.5 m) wide and is represented by lines parallel to the longitudinal centerline of the vehicle but offset 1.6 ft.

(0.5 m) from the outermost edge of the SV’s body excluding the side view mirror(s). The rearward projection begins

at the rearmost part of the SV side mirror housing and ends at a rearward boundary that is dependent on the relative

speed between the SV and POV. The blind zone is fully described in the test procedure.

of 0.25 to 0.75 m/s (0.8 to 2.5 ft./s). This test is repeated for a POV approach from both

the left and the right side of the SV.

- To pass a test trial: during the converge lane change, the BSW alert must be presented

by a time no later than 300 ms after any part of the POV enters the SV blind zone and

must remain on while any part of the POV resides within the SV blind zone; and

during the diverge lane change, the BSW alert may remain active only when the

lateral distance between the SV and POV is greater than 3 m (9.8 ft.) but less than or

equal to 6 m (19.7 ft.). The BSW alert shall not be active once the lateral distance

between the SV and POV exceeds 6 m (19.7 ft.).

 Straight Lane Pass-by Test – The POV approaches and then passes the SV while being

driven in an adjacent lane. For each trial, the SV is traveling at a constant speed of 72.4

kph (45 mph) whereas the POV is traveling at one of four constant speeds – 80.5, 88.5,

96.6, or 104.6 kph (50, 55, 60, or 65 mph). The lateral distance between the two

vehicles, defined as the closest lateral distance between adjacent sides of the polygons

used to represent each vehicle, shall nominally be 1.5 m (4.9 ft.) for the duration of the

trial. This test is repeated for a POV approach towards the SV from an adjacent lane to

the left and to the right of the SV.

- To pass a test trial, the BSW alert must be presented by a time no later than 300 ms

after the front-most part of the POV enters the SV blind zone and remain on while the

front-most part of the POV resides behind the front-most part of the SV blind zone.

The BSW alert shall not be active once the longitudinal distance between the front-

most part of the SV and the rear-most part of the POV exceeds the BSW termination

distance specified for each POV speed.

For the BSW tests, each scenario is tested using seven repeated trials for each

combination of approach direction (left and right side of the SV) and test speed. This translates

to a total of 14 tests overall for the Straight Lane Converge and Diverge Test and 56 tests overall

for the Straight Lane Pass-by Test. NCAP is proposing that to pass the NCAP system

performance requirements, the SV must pass at least five out of seven trials conducted for each

approach direction and test speed.

The proposed BSW tests represent pre-crash scenarios that correspond to a substantial

portion of fatalities and injuries observed in real-world lane change crashes. A review of

Volpe’s 2011-2015 data set showed that approximately 28 percent of fatalities and 57 percent of

injuries in lane change crashes occurred on roads with posted speeds of 72.4 kph (45 mph) or

lower.

For crashes where the travel speed was reported in FARS and GES, approximately 14

percent of fatalities and 24 percent of injuries occurred at speeds of 72.4 kph (45 mph) or

lower.

Furthermore, Volpe found that speeding was a factor in only 18 percent of the fatal lane

change crashes and 3 percent of lane change crashes that resulted in injuries. This suggests that

posted speed corresponds well to travel speed in most lane change crashes.

88,89

As noted earlier, market research conducted by Consumer Reports (CR) indicated that

BSW systems are desirable in consumer interest surveys of various ADAS technologies. In fact,

CR found not only that an overwhelming majority of vehicle owners were satisfied with BSW

technology, but also that 60 percent of them believed BSW technology had helped them avoid a

crash. However, in its study to evaluate the real-world effectiveness of ADAS technologies in

model year 2013-2017 General Motors’ (GM) vehicles, UMTRI found that GM’s Side Blind

Zone Alert produced a non-significant 3 percent reduction in lane change crashes. When the

The posted speed limit was either not reported or was unknown in 2 percent of fatal lane change crashes and 18

percent of lane change crashes that resulted in injuries.

The travel speed was either not reported or was unknown in 60 percent of fatal lane change crashes and 68 percent

of lane change crashes that resulted in injuries.

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019, August), Statistics of light-

vehicle pre-crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington,

DC: National Highway Traffic Safety Administration.

It was unknown or not reported whether speeding was a factor in 3 percent of fatal lane change crashes and 7

percent of lane change crashes that resulted in injuries.

Side Blind Zone Alert technology was combined with an earlier generation technology, GM’s

Lane Change Alert, the corresponding effectiveness increased to 26 percent.

UMTRI attributed

this increase to substantially longer vehicle detection ranges for the Lane Change Alert with Side

Blind Zone Alert system compared to GM’s earlier generation Side Blind Zone Alert system.

An Agency study of three BSW-equipped vehicles also showed that that currently available

BSW systems may likely exhibit differences in detection capabilities and operating conditions

such that their effectiveness estimates could vary significantly.

For instance, one vehicle’s

system may simply augment a driver’s visual awareness whereas another may effectively prevent

crashes by warning of higher speed lane change events. In its response to NCAP’s December

2015 notice, Bosch provided similar insight. The company stated that some BSW systems may

only provide benefit for shorter detection distances, such as 7 m (23.0 ft.) rearward, whereas

other systems may provide detection for distances up to 70 m (229.7 ft.) rearward, which would

help the driver avoid collisions with vehicles approaching from the rear in adjacent lanes at high

speeds. The Agency plans to study these performance differences in its testing.

NHTSA is proposing to conduct BSW tests in NCAP in accordance with the Agency’s

BSW test procedure. The Agency believes that the Straight Lane Pass-by Test scenario, which

stipulates incrementally higher test speeds for the POV, could be used to distinguish between

vehicles that have basic versus advanced BSW capability. For instance, an SV that can only

satisfy the BSW activation criteria when the POV approaches with a low relative velocity may

be considered as having basic BSW capability, whereas a vehicle that can look further rearward,

to sense a passing vehicle travelling at a much higher speed, may be considered to have superior

Leslie, A. J., Kiefer, R. J., Meitzner, M. R., & Flannagan, C. A. (2019), Analysis of the field effectiveness of

General Motors production active safety and advanced headlighting systems, The University of Michigan

Transportation Research Institute and General Motors LLC, UMTRI-2019-6.

For GM’s Lane Chane Alert systems, sensors in the vehicle’s rear bumper are utilized to warn the driver of

vehicles approaching from the rear on either the left or right side.

Forkenbrock, G., Hoover, R. L., Gerdus, E., Van Buskirk, T. R., & Heitz, M. (2014, July), Blind spot monitoring

in light vehicles—System performance (Report No. DOT HS 812 045), Washington, DC: National Highway Traffic

Safety Administration.

BSW capability. NHTSA believes such an assessment is important because when one vehicle

encroaches into the adjacent lane of the other, the crashes associated with higher speed

differentials can be expected to be more severe than those that occur when the two vehicle

speeds are more similar. Furthermore, the capability of a vehicle to detect when another vehicle

has entered an extended rear zone could be important for the application of other ADAS

technologies such as blind spot intervention (BSI) or SAE

Level 2 partial driving automation

systems that incorporate automatic lane change features. Therefore, the Agency believes that

long-range vehicle detection may not only increase the effectiveness of blind spot technologies

such as BSI, but also enhance capabilities and robustness of other ADAS applications. For these

reasons, NHTSA is proposing (later in this notice) the incorporation of BSI technology in NCAP

to encourage the proliferation of such systems along with sensing strategies that offer a greater

field of view.

Commenters to NHTSA’s December 2015 notice overwhelmingly supported the addition

of BSW in NCAP. In fact, many commenters suggested the Agency expand the testing

requirements to encompass additional test targets, such as motorcycles, and test conditions.

Several commenters also recommended that NHTSA harmonize its BSW test procedure with

International Organization for Standardization (ISO) standards. Each of these topics will be

discussed below.

a. Additional Test Targets and/or Test Conditions

Commenters, including the ASC, Continental, Bosch, NSC, and others, recommended

that the Agency expand the BSW testing requirements to include motorcycle detection. Delphi,

MTS, Medical College of Wisconsin (MCW), and CU suggested that NHTSA evaluate a

SAE International (2018), SAE J3016_201806: Taxonomy and definitions for terms related to driving automation

systems for on-road motor vehicles, Warrendale, PA, www.sae.org.

The sustained driving automation system of both the lateral and longitudinal vehicle motion control with the

expectation that the driver supervises the driving automation system.

vehicle’s ability to detect bicycles in addition to motorcycles. Similarly, Subaru suggested that

changes to the Straight Lane Pass-by Test should be made to address motorcycle detection. MTS

and MCW added that motorcycle riders and bicyclists are more vulnerable to serious and fatal

injuries compared to occupants of motor vehicles. A few commenters were not supportive of

adding a motorcycle detection test in NCAP. Global Automakers and Hyundai stated that

although it was a reasonable goal for the future, no standardized test devices currently existed at

the time. Similarly, Honda and the Alliance recommended that the Agency focus on vehicle

detection as a first step since no standard test procedure exists for motorcycle detection. The

Alliance added that since the location of a motorcycle within a lane can vary greatly, test

procedures would need to specify motorcycle behavior and reasonable detection distances.

Furthermore, MTS stated that the position of the motorcycle POV within the lane (near, center,

far) should be specified, and the radar cross section and projected area of the motorcycle should

be considered as well.

NHTSA agrees that BSW systems capable of detecting motorcycles would improve

safety. A review of the 2011 through 2015 FARS and GES data sets

showed that there were

106 fatal crashes and nearly 5,100 police-reported crashes annually, on average, for same

direction lane change crashes involving a vehicle and motorcycle. In comparison, as mentioned

earlier, there were 542 fatalities and 503,070 police-reported crashes annually, on average, for

lane change crashes involving motor vehicles. These data show that more occupants of motor

vehicles die in lane changing crashes than do motorcyclists. However, the fatality rate for

motorcyclists is greater than that for vehicle occupants.

At this time, the Agency has decided to prioritize testing of BSW systems on motor

vehicles for NCAP. NHTSA believes that performing BSW testing on light vehicles,

Swanson, E., Azeredo, P., Yanagisawa, M., & Najm, W. (2018, September), Pre-Crash Scenario Characteristics of

Motorcycle Crashes for Crash Avoidance Research (Report No. DOT HS 812 902), Washington, DC: National

Highway Traffic Safety Administration. In Press

particularly at higher POV closing speeds, and for active safety systems (as will be discussed

next), should encourage development of robust sensing systems, which may improve the

detection of other objects such as motorcycles. That being said, the Agency has planned an

upcoming research project designed to address injuries and fatalities for other vulnerable road

users, specifically motorcyclists. The Agency will continue to observe the development of BSW

technology and is likely to include test procedures for motorcycle detection in NCAP at a later

date if the technology meets the four prerequisites mentioned above.

Several commenters offered additional suggestions for ways NHTSA could expand the

BSW test procedure. MCW suggested that the Agency adopt test scenarios that address curved

roads and low light conditions. CU proposed that the Agency should assess whether BSW

systems provide a clear indication to the driver that the system is not operating since sensors are

sometimes rendered inoperable in poor weather or when blocked.

As with all the ADAS technologies, NHTSA recognizes that there is a need to understand

and assure crash mitigation performance of BSW systems under all practical situations that the

driver and vehicle will encounter in the real world. However, such comprehensive testing is not

always practical within the scope of the NCAP program. Thus, for technologies that met the four

principles for inclusion in NCAP, the Agency primarily attempted to address the most frequently

occurring, most fatal, and most injurious pre-crash scenarios when prioritizing tests to add to the

program. When ADAS technologies penetrate the fleet in sufficient numbers, then the Agency

can evaluate how these systems are performing in the real world and adjust the system

performance criteria accordingly to address additional test conditions, such as those mentioned

by MCW. Regarding CU’s suggestion, the Agency believes, after reviewing vehicle owner’s

manuals, that most vehicle manufacturers are including provisions in their system designs to

provide a malfunction indicator to the driver if the system is no longer operational because the

sensors are blocked or due to severe weather conditions.

NHTSA has also considered Bosch’s request to expand the definition of BSW to

encourage adoption of systems that provide longer detection distances. NHTSA believes, as

discussed above, that by using higher POV closing speeds to assess BSW system performance, it

may effectively drive enhanced blind spot system capabilities such as those required for other

rearward-looking ADAS applications, like BSI, or automatic lane change functions.

b. Test Procedure Harmonization

Several commenters suggested that NHTSA harmonize its BSW test procedure with

International Organization for Standardization (ISO) standard 17387:2008, Intelligent transport

systems—Lane change decision aid systems (LCDAS)— Performance requirements and test

procedures or with various aspects of this standard. Global Automakers and Hyundai

commented that NHTSA should shift the forward edge of the blind zone rearward from the

outside rearview mirrors to the eye point of a 95

percentile person, as specified in ISO 17387.

Hyundai stated that the ISO procedure is designed such that when the POV is in-line with the SV

driver’s eye ellipse, the driver’s peripheral vision allows him/her to see the POV without the

assistance of BSW systems. The ASC, Continental, and Subaru also suggested that the Agency

align the warning zones in the Agency’s BSW test procedure with those specified in ISO 17387.

The Agency does not agree with commenters’ suggestion to adopt the ISO procedure for

defining the forward edge of the blind zone as measured using the eye ellipse from a seated 95

percentile person. NHTSA believes that the blind zone should be defined not by a specific

seated individual but by the vehicle’s characteristics, since a real-world blind spot for any

particular vehicle would differ depending on the size characteristics of the individual driving the

vehicle at the time. Since people vary in size, they will sit in different seating positions and have

different seating preferences. For instance, a 95

percentile male will be seated more rearward

whereas a 5

percentile female will be seated more forward. In addition, drivers have personal

preferences for adjusting their side view mirrors that may not be considered optimal and may not

provide a full field of view when checking the mirrors to make change lanes. For these reasons,

the Agency tentatively concludes that it is more appropriate and better for the safety of

consumers to set the forward plane of the blind zone at the rearmost part of the side view

mirrors, as specified in its BSW test procedure. This approach should not only best

accommodate a wide variety of driver sizes and seating positions, but also reduce test complexity

when defining the blind zone.

2. Adding Blind Spot Intervention (BSI)

Blind spot intervention (BSI) systems are similar to AEB and LKS systems in that they

provide active intervention to help the driver avoid a collision with another vehicle. BSW

systems alert a driver that a vehicle is in his/her blind spot, whereas BSI systems activate when

the BSW alert is ignored, and intervene either by automatically applying the vehicle’s brakes or

providing a steering input to guide the vehicle back into the unobstructed lane. With their active

capability, BSI systems can help a driver avoid collisions with other vehicles that are

approaching the vehicle’s blind spot, in addition to preventing crashes with vehicles operating

within the vehicle’s blind spot.

Like BSW systems, BSI systems utilize rear-facing sensors to detect other vehicles that

are next to or behind the vehicle in adjacent lanes. Depending on the design of these systems,

BSI activation may or may not require the driver to operate his/her turn signal indicator during a

lane change. Furthermore, some BSI systems may only operate if the vehicle’s BSW system is

also enabled.

As discussed earlier, UMTRI found that GM’s BSW system, Side Blind Zone Alert,

produced a non-significant 3 percent reduction in lane change crashes. However, when Side

Blind Zone Alert was combined with a later generation technology, GM’s Lane Change Alert,

the corresponding effectiveness increased to 26 percent.

Given BSI is only now penetrating the

fleet, NHTSA is unaware of any effectiveness studies for this technology. However, as

discussed earlier, the Agency believes that active safety technologies are more effective than

warning technologies. The UMTRI study concluded that AEB is more effective than FCW alone

and that LKS is more effective than LDW. The Agency believes the same relationship will

likely hold true for blind spot systems, and that BSI will be more effective than BSW alone.

NHTSA also believes, as mentioned above, that adopting ADAS technologies such as BSI

should also encourage development of enhanced BSW system capabilities (e.g., motorcycle and

bicycle detection), and may increase the robustness of other ADAS applications.

NHTSA is proposing to use its published draft test procedure titled, “Blind Spot

Intervention System Confirmation Test,”

to evaluate the performance of vehicles equipped with

BSI technology in NCAP. The Agency’s test procedure consists of three scenarios: Subject

Vehicle (SV) Lane Change with Constant Headway, SV Lane Change with Closing Headway,

and SV Lane Change with Constant Headway, False Positive Assessment. In the first two

scenarios, an SV initiates or attempts a lane change into an adjacent lane while a single POV is

residing within the SV’s blind zone (Scenario 1), or is approaching it from the rear (Scenario 2).

The third scenario is used to evaluate the propensity of a BSI system to activate inappropriately

in a non-critical driving scenario that does not present a safety risk to the occupants in the SV. In

each of the tests, the POV is a strikeable object with the characteristics of a compact passenger

car. The system performance requirements stipulate that the SV may not contact the POV during

the conduct of any test trial. NHTSA is requesting comment on the number of trials that are

Leslie, A. J., Kiefer, R. J., Meitzner, M. R., & Flannagan, C. A. (2019), Analysis of the field effectiveness of

General Motors production active safety and advanced headlighting systems, The University of Michigan

Transportation Research Institute and General Motors LLC, UMTRI-2019-6.

84 FR 64405 (Nov. 21, 2019).

appropriate for each test. Each of these scenarios, along with the proposed evaluation criteria, is

detailed below:

 SV Lane Change with Constant Headway – The POV is driven at 72.4 kph (45 mph) in a

lane adjacent and to the left of the SV also traveling at 72.4 kph (45 mph) with a constant

longitudinal offset such that the front-most part of the POV is 1 m (3.3 ft.) ahead of the

rear-most part of the SV. After a short period of steady-state driving, the SV driver

engages the left turn signal indicator at least 3 s after all pre-SV lane change test validity

criteria have been satisfied. Within 1.0 ± 0.5 s after the turn signal has been activated,

the SV driver initiates a manual lane change into the POV’s travel lane. The SV driver

then releases the steering wheel within 250 ms of the SV exiting a 800.1 m (2,625 ft.)

radius curve during the lane change. To meet the performance criteria, the BSI system

must intervene so as to prevent the left rear of the SV from contacting the right front of

the POV. Additionally, the SV BSI intervention shall not cause the SV to travel 1.0 ft.

(0.3 m) or more beyond the inboard edge of the lane line separating the SV travel lane

from the lane adjacent and to the right of it within the validity period.

 SV Lane Change with Closing Headway Scenario – The POV is driven at a constant

speed of 80.5 kph (50 mph) towards the rear of the SV in an adjacent lane to the left of

the SV, which is traveling at a constant speed of 72.4 kph (45 mph). During the test, the

SV driver engages the turn signal indicator when the POV is 4.9 ± 0.5 s from a vertical

plane defined by the rear of the SV and perpendicular to the SV travel lane. Within 1.0 ±

0.5 s after the turn signal has been activated, the SV driver initiates a manual lane change

into the POV’s travel lane. The SV driver then releases the steering wheel within 250 ms

The Agency notes that these test scenario descriptions assume the SV is operating in SAE Automation Level 0 or

Level 1 operation with only the Automatic Cruise Control (ACC) enabled. Though the Agency’s BSI test procedure

has provisions to evaluate vehicles operating in SAE Automation Levels 2 or 3. Test scenario descriptions for these

evaluations are not discussed herein.

of the SV exiting a 800.1 m (2,625 ft.) radius curve. To meet the performance criteria,

the BSI system must intervene to prevent the left rear of the SV from contacting the right

front of the POV. Additionally, the SV BSI intervention shall not cause the SV to travel

1.0 ft. (0.3 m) or more beyond the inboard edge of the lane line separating the SV travel

lane from the lane adjacent and to the right of it within the validity period.

 SV Lane Change with Constant Headway, False Positive Assessment Test – The POV is

driven at 72.4 kph (45 mph) in a lane that is two lanes to the left of the SV’s initial travel

lane with a constant longitudinal offset such that the front-most part of the POV is 1 m

(3.3 ft.) ahead of the rear-most part of the SV, which is also travelling at 72.4 kph (45

mph). The SV driver engages the left turn signal indicator at least 3 s after all pre-SV

lane change test validity criteria have been satisfied. Within 1.0 ± 0.5 s after the turn

signal has been activated, the SV driver initiates a manual lane change into the left

adjacent lane (the one between the SV and POV). For this test, the driver does not

release the steering wheel. Since the lane change will not result in an SV-to-POV impact,

the SV BSI system must not intervene during any valid trials. To determine whether a

BSI intervention occurred, the SV yaw rate data collected during the individual trials

performed in this scenario are compared to a baseline composite. After being aligned in

time to the baseline, the difference between the data must not exceed 1 degree/second

within the test validity period.

The proposed crash-imminent BSI test scenarios represent pre-crash scenarios that

correspond to a substantial portion of fatalities and injuries observed in real-world lane change

crashes. As discussed in the BSW crash statistics section, Volpe showed that approximately 28

percent of fatalities and 57 percent of injuries in lane change crashes occurred on roads with

posted speeds of 72.4 kph (45 mph) or lower.

Furthermore, approximately 14 percent of

fatalities and 24 percent of injuries were reported for crashes that occurred at known travel

speeds of 72.4 kph (45 mph) or lower.

100

NHTSA has conducted a series of tests utilizing its proposed BSI test procedure. Since

BSI systems are not widely available in the fleet, the Agency selected vehicles in order to cover

as many manufacturers as possible that have implemented this technology. All vehicles selected

for BSW testing also underwent BSI testing. Test reports related to both test programs can be

found in the docket for this notice. For the purposes of this testing, the Agency used the Global

Vehicle Target (GVT) Revision G to represent the POV, which is specified in the BSI test

procedure as a strikeable object.

101

When the BSI technology assessment is incorporated into

NCAP, the Agency plans to use the GVT Revision G as a strikeable target to be consistent with

Euro NCAP’s ADAS test procedures that specify a strikeable target. In the context of testing

BSW and BSI technologies in NCAP to address lane change crashes, NHTSA is seeking

comment on the following:

(16) Should all BSW testing be conducted without the turn signal indicator activated? Why

or why not? If the Agency was to modify the BSW test procedure to stipulate

activation of the turn signal indicator, should the test vehicle be required to provide an

audible or haptic warning that another vehicle is in its blind zone, or is a visual warning

sufficient? If a visual warning is sufficient, should it continually flash, at a minimum,

to provide a distinction from the blind spot status when the turn signal is not in use?

Why or why not?

The posted speed limit was either not reported or was unknown in 2 percent of fatal lane change crashes and 18

percent of lane change crashes that resulted in injuries.

100

The travel speed was either not reported or was unknown in 65 percent of fatal lane change crashes and 67

percent of lane change crashes that resulted in injuries.

101

The GVT is a three-dimensional surrogate that resembles a white hatchback passenger car. It is currently used by

other consumer organizations, including Euro NCAP, and vehicle manufacturers in their internal testing of ADAS

technologies. See Section III.D.2. of this notice for an expanded discussion of the GVT.

(17) Is it appropriate for the Agency to use the Straight Lane Pass-by Test to quantify and

ultimately differentiate a vehicle’s BSW capability based on its ability to provide

acceptable warnings when the POV has entered the SV’s blind spot (as defined by the

blind zone) for varying POV-SV speed differentials? Why or why not?

(18) Is using the GVT as the strikeable POV in the BSI test procedure appropriate? Is using

Revision G in NCAP appropriate? Why or why not?

(19) The Agency recognizes that the BSW test procedure currently contains two test

scenarios that have multiple test conditions (e.g., test speeds and POV approach

directions (left and right side of the SV)). Is it necessary for the Agency to perform all

test scenarios and test conditions to address the real-world safety problem adequately,

or could it test only certain scenarios or conditions to minimize test burden in NCAP?

For instance, should the Agency consider incorporating only the most challenging test

conditions into NCAP, such as the ones with the greatest speed differential, or choose

to perform the test conditions having the lowest and highest speeds? Should the

Agency consider only performing the test conditions where the POV passes by the SV

on the left side if the vehicle manufacturer provides test data to assure the left side pass-

by tests are also representative of system performance during right side pass-by tests?

Why or why not?

(20) Given the Agency’s concern about the amount of system performance testing under

consideration in this RFC, it seeks input on whether to include a BSI false positive test.

Is a false positive assessment needed to insure system robustness and high customer

satisfaction? Why or why not?

(21) The BSW test procedure includes 7 repeated trials for each test condition (i.e., test

speed and POV approach direction). Is this an appropriate number of repeat trials?

Why or why not? What is the appropriate number of test trials to adopt for each BSI

test scenario, and why? Also, what is an appropriate pass rate for each of the two tests,

BSW and BSI, and why is it appropriate?

(22) Is it reasonable to perform only BSI tests in conjunction with activation of the turn

signal? Why or why not? If the turn signal is not used, how can the operation of BSI

be differentiated from the heading adjustments resulting from an LKS intervention?

Should the SV’s LKS system be switched off during conduct of the Agency’s BSI

evaluations? Why or why not?

C. Adding Pedestrian Automatic Emergency Braking (PAEB)

Another important ADAS technology NHTSA proposes to include in its upgrade of

NCAP is pedestrian automatic emergency braking (PAEB). PAEB systems function similar to

AEB systems but detect pedestrians instead of vehicles. PAEB uses information from forward-

looking sensors to issue a warning and actively apply the vehicle’s brakes when a pedestrian, or

sometimes a cyclist, is in front of the vehicle and the driver has not acted to avoid the impending

impact. Similar to AEB, PAEB systems typically use cameras to determine whether a pedestrian

is in imminent danger of being struck by the vehicle, but some systems may use a combination of

cameras, radar, lidar, and/or thermal imaging sensors.

Many pedestrian crashes occur when a pedestrian is in the forward path of a driver’s

vehicle. Four common pedestrian crash scenarios include when the vehicle is:

1. Heading straight and a pedestrian is crossing the road;

2. Turning right and a pedestrian is crossing the road;

3. Turning left and a pedestrian is crossing the road; and

4. Heading straight and a pedestrian is walking along or against traffic.

These four crash scenarios are defined as Scenarios S1-S4, respectively, by the Crash

Avoidance Metrics Partnership (CAMP) Crash Imminent Braking (CIB) Consortium.

102

Two of these scenarios, S1 and S4, are included in NHTSA’s draft research PAEB test

procedure, published on November 21, 2019, and referenced herein as the 2019 PAEB test

procedure.

103

The S1 scenario represents a pedestrian crossing the road in front of the vehicle,

while the S4 scenario represents a pedestrian moving with or against traffic along the side of the

road in the path of the vehicle. Both test scenarios are repeated for multiple pedestrian impact

locations. The S1 and S4 crash scenarios were chosen for inclusion in NHTSA’s 2019 PAEB

test procedure because a review of pedestrian crashes from the 2011 through 2012 GES and

FARS data sets

104

found that, on average, these two pre-crash scenarios (S1 and S4) accounted

for approximately 33,000 (52 percent) of vehicle-pedestrian crashes and 3,000 (90 percent) fatal

vehicle-pedestrian crashes with a light-vehicle striking a pedestrian as the first event.

Furthermore, these crashes accounted for 67 percent of MAIS 2+ and 76 percent of MAIS 3+

injured pedestrians.

105

The 2019 PAEB test procedure only considered daylight test conditions

for both the S1 and S4 crash scenarios.

The Agency’s 2019 PAEB test procedure does not include CAMP scenario S2 (vehicle

turning right and a pedestrian crossing the road), and CAMP scenario S3 (vehicle turning left and

a pedestrian crossing the road). In response to the December 2015 notice, several commenters

stated that addressing these scenarios with available technology may generate a significant

102

Carpenter, M. G., Moury, M. T., Skvarce, J. R., Struck, M. Zwicky, T. D., & Kiger, S. M. (2014, June), Objective

tests for forward looking pedestrian crash avoidance/mitigation systems: Final report (Report No. DOT HS 812

040), Washington, DC: National Highway Traffic Safety Administration.

103

84 FR 64405 (Nov. 21, 2019).

104

Yanagisawa, M., Swanson, E., Azeredo, P., & Najm, W. G. (2017, April), Estimation of potential safety benefits

for pedestrian crash avoidance/mitigation systems (Report No. DOT HS 812 400), Washington, DC: National

Highway Traffic Safety Administration.

105

As explained previously, the Abbreviated Injury Scale (AIS) is a classification system for assessing impact injury

severity. AIS ranks individual injuries by body region on a scale of 1 to 6 where 1=minor, 2=moderate, 3=serious,

4=severe, 5=critical, and 6=maximum (untreatable). MAIS represents the maximum injury severity, or AIS level,

recorded for an occupant (i.e., the highest single AIS for a person with one or more injuries).

number of false positive detections. Such false detections could have the unintended

consequences of causing hazardous situations (e.g.., unexpected sudden braking while turning in

traffic) that could lead drivers to disable their PAEB systems, or even lead to an increase in rear-

end collisions. The commenters explained that the S2 and S3 test scenarios require more

sophisticated algorithms as well as more robust test methodologies than those required for

scenarios S1 and S4. However, ZF TRW mentioned that ADAS sensors designed to meet Euro

NCAP’s Vulnerable Road Users test procedures would have increased fields of view (FOV),

which should improve their effectiveness in turning scenarios. Others stated that the articulating

mannequins may not be representative of a real human for all sensing technologies in turning

scenarios. Most commenters indicated that it was more appropriate to focus on the scenarios

affording the most significant safety benefits first – S1 and S4. Commenters stated that adding

the S2 and S3 scenarios would be more practical when the technology matures. NHTSA will

continue to evaluate PAEB systems to assess the feasibility of expanding the suite of PAEB tests

as technological advancements are made. The Agency will consider adding these test scenarios

(S2 and S3) to NCAP in the future once the Agency has repeatable and reliable test data to

support their inclusion.

In the 2019 PAEB test procedure, the S1 test scenario includes seven different test

conditions—S1a, S1b, S1c, S1d, S1e, S1f, and S1g. For these tests, the SV travels in a straight,

forward direction at 40 kph (24.9 mph). Additionally, the SV also travels at 16 kph (9.9 mph)

for test conditions S1a, S1b, S1c, and S1d. A pedestrian mannequin crosses perpendicular to the

subject vehicle’s line of travel at 5 kph (3.1 mph) for all test conditions, except for S1e, in which

the mannequin crosses at 8 kph (5.0 mph). In test condition S1a, the SV encounters a crossing

adult pedestrian mannequin walking from the nearside (i.e., the passenger’s side of the vehicle)

with 25 percent overlap of the vehicle.

106

In test conditions S1b and S1c, the SV encounters a

crossing adult pedestrian walking from the nearside with 50 percent and 75 percent overlap of

the vehicle, respectively. In test condition S1d, the SV encounters a crossing child pedestrian

mannequin running from behind parked vehicles from the nearside with 50 percent overlap of the

vehicle. In test condition S1e, the SV encounters a crossing adult pedestrian running from the

“offside” (i.e., the driver’s side of the vehicle) with 50 percent overlap of the vehicle. In test

condition S1f, the SV encounters a crossing adult pedestrian walking from the nearside that stops

short (-25% overlap) of entering the vehicle’s path. In test condition S1g, the SV encounters a

crossing adult pedestrian walking from the nearside that clears the vehicle’s path (125%

overlap).

The S4 test scenario in the 2019 PAEB test procedure includes three different test

conditions—S4a, S4b, and S4c. In this test scenario, the SV travels in a straight, forward

direction at 40 kph (24.9 mph) and/or 16 kph (9.9 mph) (for test conditions S4a and S4b) and a

pedestrian mannequin moves parallel to the flow of traffic at 5 kph (3.1 mph) (for test condition

S4c) or is stationary (for test condition S4a and S4b) in front of the SV. For all S4 test

conditions, the SV is aligned to impact the pedestrian at 25 percent overlap. In test condition

S4a, the SV encounters an adult pedestrian standing in front of the vehicle on the nearside of the

road facing away from the approaching SV. In test condition S4b, the SV encounters an adult

pedestrian standing in front of the vehicle on the nearside of the road facing towards the

approaching SV. In test condition S4c, the SV encounters an adult pedestrian walking in front of

the vehicle on the nearside of the road facing away from the approaching SV.

The Agency is proposing to make several changes to the 2019 PAEB test procedure for

the purpose of adopting it for use in NCAP. These changes involve the pedestrian mannequins,

106

Overlap is defined as the percent of the vehicle’s width that the pedestrian would traverse prior to impact if the

vehicle’s speed and pedestrian’s speed remain constant.

test speeds and included test conditions, the specified lighting conditions, and the number of test

trials required to be conducted for each test condition.

The first change the Agency is proposing to make to the 2019 PAEB test procedure

concerns the pedestrian targets. As was recommended by several commenters who responded to

the December 2015 notice, the Agency proposes to utilize state-of-the-art mannequins with

articulated, moving legs, instead of the posable child and adult pedestrian test mannequins

specified in the 2019 PAEB test procedure. NHTSA believes that the articulating pedestrian

targets are more representative of walking pedestrians and expects that these more realistic

targets will encourage development of PAEB systems that detect, classify, and respond to

pedestrians more accurately and effectively. In turn, this should allow manufacturers to improve

the effectiveness of current PAEB systems. The Agency also recognizes that adopting the child

and adult articulating targets would harmonize with other major consumer information-focused

entities that use articulating mannequins, such as Euro NCAP and IIHS. The Bipartisan

Infrastructure Law mandated that NHTSA identify opportunities where NCAP would “benefit

from harmonization with third-party safety rating programs,” and the Agency believes that the

pedestrian mannequins represent one such opportunity.

The second change the Agency is proposing to make to the 2019 PAEB test procedure for

incorporation into NCAP involves test speeds. The test speeds specified in the 2019 PAEB test

procedure correspond to a relatively small percentage of crashes that result in pedestrian injuries

and fatalities. Volpe’s analysis of 2011-2015 FARS and GES crash data sets showed that 9

percent of pedestrian fatalities and 25 percent of pedestrian injuries resulted from crashes that

occurred on roadways with posted speeds of 40.2 kph (25 mph) or less, whereas 88 percent of

fatalities and 43 percent of injuries occurred for crashes on roadways with posted speeds greater

than 40.2 kph (25 mph).

107,108

For crashes that occurred on roadways where the travel speed was

known, 6 percent of pedestrian fatalities and 19 percent of pedestrian injuries were reported for

travel speeds of 40.2 kph (25 mph) or less, whereas 36 percent of fatalities and 7 percent of

injuries occurred for travel speeds greater than 40.2 kph (25 mph).

109

NHTSA notes that

speeding was a factor in only 5 percent of the fatal pedestrian crashes, which suggests that the

posted speed could correlate closely with the travel speed of the vehicle prior to impact with the

pedestrian.

110,111

As Volpe’s analysis focused on 2011-2015 FARS and GES crash data sets, it is likely

that most vehicles studied were not equipped with PAEB systems. Recently, IIHS studied

approximately 1,500 police-reported crashes involving a wide variety of 2017-2020 model year

vehicles from various manufacturers to examine the effects of PAEB systems on real-world

pedestrian crashes.

112

In this study, the Institute found that “pedestrian AEB was associated with

a 32 percent reduction in the odds of a pedestrian crash on roads with speed limits of 25 mph or

less and a 34 percent reduction on roads with 30-35 mph limits, but no reduction at all on roads

with speed limits of 50 mph or higher…”. These findings highlight the limitations of existing

PAEB systems and the importance of adopting higher test speeds for PAEB testing (where

feasible) to encourage additional safety improvement.

To establish feasible speed thresholds for adoption in its PAEB test procedure, the

Agency conducted a series of tests on a selection of MY 2020 vehicles from various

107

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019, August), Statistics of light-

vehicle pre-crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington,

DC: National Highway Traffic Safety Administration.

108

The posted speed limit was either not reported or was unknown in 4 percent of fatal pedestrian crashes and 29

percent of pedestrian crashes that resulted in injuries.

109

The travel speed was either not reported or was unknown in 59 percent of fatal pedestrian crashes and 72 percent

of pedestrian crashes that resulted in injuries.

110

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019, August), Statistics of light-

vehicle pre-crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington,

DC: National Highway Traffic Safety Administration.

111

In 4 percent of pedestrian crashes, it was unknown or not reported whether speeding was a factor.

112

Cicchino, J. B (2022, February), Effects of automatic emergency braking systems on pedestrian crash risk,

Insurance Institute for Highway Safety, https://www.iihs.org/api/datastoredocument/bibliography/2243.

manufacturers to assess the operational range and performance of current PAEB systems.

Vehicles for the PAEB characterization tests were selected with the intent of testing a variety of

vehicle makes, types, sizes; global and domestic products; and forward-facing sensor types

(camera only, stereo camera, fused camera plus radar, etc.) for a given manufacturer and across

all manufacturers.

For the purpose of this study, the Agency used the 2019 PAEB test procedure, but

employed the articulating mannequins in lieu of the posable mannequins and expanded the test

procedure specifications to include increased vehicle test speeds for the S1b, S1d, S1e, S4a, and

S4c test conditions. For these tests, the SV speed was incrementally increased to identify when

each SV reached its operational limits and did not respond to the pedestrian target. Before the

tests were initiated, the maximum test speeds for the S1 and S4 scenarios were set to 60 kph

(37.2 mph) and 80 kph (49.7 mph), respectively.

113

These maximum speeds are consistent with

Euro NCAP’s AEB Vulnerable Road User test protocol and correspond to up to 74 percent of

fatal pedestrian crashes and 65 percent of injurious pedestrian crashes that occurred on U.S.

roadways, per Volpe’s 2011-2015 FARS and GES analysis of posted speed data.

114

When no or

late intervention occurred for a vehicle and test condition (i.e., combination of test scenario and

speed), NHTSA repeated the test condition at a test speed that was 5 kph (3.1 mph) lower. This

reduced speed defined the system’s upper capabilities.

A test matrix of the PAEB characterization study regarding test speed is provided below.

 Full PAEB test series (includes S1 a-g and S4 a-c)

Daytime light conditions, articulating dummies, and additional SV test speeds in kph

(mph) for S1b, d, and e, and S4a and c, as shown in Table 4.

113

These test speeds represent the maximum test speeds potentially utilized for a given test condition. The actual

speeds used for a given combination of vehicle and test condition depended on observed PAEB system performance.

114

European New Car Assessment Programme (Euro NCAP). (2019, July). TEST PROTOCOL – AEB VRU systems

3.0.2.

Table 4: Complete matrix of the PAEB characterization study

Scenario S1a S1b S1c S1d S1e S1f S1g S4a S4b S4c

Subject

Vehicle

Speed

(kph/

mph)

16.0/

9.9

16.0/

9.9

16.0/

9.9

16.0/

9.9

40.0/

24.9

40.0/

24.9

40.0/

24.9

16.0/

9.9

16.0/

9.9

16.0/

9.9

40.0/

24.9

20.0/

12.4

40.0/

24.9

20.0/

12.4

50.0/

31.1

40.0/

24.9

40.0/

24.9

40.0/

24.9

30.0/

18.6

30.0/

18.6

60.0/

37.3

50.0/

31.1

50.0/

31.1

40.0/

24.9

40.0/

24.9

60.0/

37.3

60.0/

37.3

50.0/

31.1

50.0/

31.1

70.0/

43.5

70.0/

43.5

60.0/

37.3

60.0/

37.3

80.0/

49.7

80.0/

49.7

The Agency’s characterization testing showed that many MY 2020 vehicles were able to

repeatedly avoid impacting the pedestrian mannequins at higher test speeds than those specified

in the 2019 PAEB test procedure. In fact, several vehicles repeatably achieved full crash

avoidance at speeds up to 60 kph (37.3 mph) or higher for the assessed S1 and S4 test conditions.

Test reports related to this testing can be found in the docket for this notice.

In light of these results, NHTSA is proposing to increase the maximum SV test speed

from the 40 kph (24.9 mph) specified in the 2019 PAEB test procedure to 60 kph (37.3 mph) for

all PAEB test conditions the Agency is proposing to include in NCAP. These include S1a-e and

S4a-c. The Agency notes that it is not proposing to include PAEB false positive test conditions

(i.e., S1f and S1g) in NCAP at this time, but is requesting comment on whether the omission of

these test conditions is appropriate. NHTSA also notes that 60 kph (37.3 mph) is the maximum

vehicle speed Euro NCAP uses to assess PAEB performance for test conditions that are similar

to, if not identical to, some of those proposed for use in NCAP, namely S1a, c, d, and e, and S4c.

Adopting this higher test speed will also drive improved PAEB system performance to address a

larger portion of real-world fatalities and injuries.

The Agency is also proposing a minimum test speed of 10 kph (6.2 mph) for all of the

proposed test scenarios. Although this speed is lower than the minimum test speed used in the

2019 PAEB test procedure and in its characterization testing (i.e., 16 kph (9.9 mph)), it is the

minimum test speed specified in Euro NCAP’s pedestrian tests, with the exception of Euro

NCAP’s Car-to-Pedestrian Longitudinal Adult (CPLA) scenario. The minimum vehicle test

speed for the CPLA scenario, which is similar to the Agency’s PAEB S4c test scenario, is 20 kph

(12.4 mph).

115

As stated earlier, in accordance with the Bipartisan Infrastructure Law, the

Agency is taking steps to harmonize with existing consumer information rating programs where

possible and when appropriate. NHTSA also believes that reducing the minimum test speed to

10 kph (6.2 mph) will assure PAEB system functionality for crashes that may still cause injuries.

In an effort to harmonize with other consumer information programs on vehicle safety,

NHTSA is also proposing to adopt Euro NCAP’s approach to assessing vehicles’ PAEB system

performance by incrementally increasing the SV speed from the minimum test speed for a given

scenario to the maximum. The Agency is proposing 10 kph (6.2 mph) increments for this

progression in test speed. In their comments to the December 2015 notice, Global Automakers

and Mobileye encouraged NHTSA to expand the applicability of the PAEB tests, particularly the

S1 scenario, to include a broader range of test speeds because pedestrian injuries occurred over a

wide range of crash speeds, as the Agency has also indicated. The organizations also mentioned

that PAEB system performance reflects a trade-off between FOV and collision speed/detection

distance. Systems that have a narrow FOV are more effective at addressing higher speed crashes

since they can see further, and systems that have a wider FOV are more effective at addressing

lower speed impacts.

115

One difference in the Agency’s proposed S4c test condition and Euro NCAP’s CPLA test condition is the amount

of pedestrian overlap with the vehicle at the lower speed (NHTSA uses a 25 percent overlap while a 50 percent

overlap is used in Euro NCAP’s CPLA test). NHTSA believes that for the 25 percent overlap condition in S4c, a

minimum test speed of 10 kph (6.2 mph) is appropriate and does not see a reason to deviate from the minimum test

speed (10 kph (6.2 mph)) proposed for the other PAEB test conditions.

As its third change to the 2019 PAEB test procedure, the Agency is proposing to expand

PAEB evaluation to include different lighting conditions. NHTSA’s PAEB characterization

study included performance assessments for dark lighting conditions (i.e., nighttime testing), in

addition to the daylight conditions specified in the 2019 PAEB test procedure, for the same test

vehicles. For each vehicle model tested, one set of tests was conducted with the pedestrian

mannequin illuminated only by the vehicle’s lower beams and a second set of tests with the

pedestrian mannequin illuminated by the upper beams. The area where the mannequin was

located was not provided any additional (i.e., external) light source. This repeat testing was

conducted because Volpe’s 2011-2015 FARS data set showed that 36 percent of pedestrian

fatalities occurred in the dark with no overhead lights. Test matrices of the PAEB

characterization study with respect to dark lighting conditions are provided in Tables 5 and 6.

 PAEB test series (includes S1b, d, and e, and S4a and c)

Dark conditions with lower beams, articulating dummies, and additional SV test speeds

in kph (mph), are shown in Table 5.

Table 5: PAEB test series for dark conditions with lower beams

Scenario S1b S1d S1e S4a S4c

Subject Vehicle

Speed (kph/

mph)

16.0/

9.9

16.0/

9.9

40.0/

24.9

16.0/

9.9

16.0/

9.9

20.0/

12.4

20.0/

12.4

50.0/

31.1

40.0/

24.9

40.0/

24.9

30.0/

18.6

30.0/

18.6

60.0/

37.3

50.0/

31.1

50.0/

31.1

40.0/

24.9

40.0/

24.9

60.0/

37.3

60.0/

37.3

50.0/

31.1

50.0/

31.1

70.0/

43.5

70.0/

43.5

60.0/

37.3

60.0/

37.3

80.0/

49.7

80.0/

49.7

 PAEB test series (includes S1b, d, and e, and S4a and c)

Dark conditions with upper beams, articulating dummies, and additional SV test speeds

in kph (mph), are shown in Table 6.

Table 6: PAEB test series for dark conditions with upper beams

Scenario S1b S1d S1e S4a S4c

Subject Vehicle

Speed (kph/

mph)

16.0/

9.9

16.0/

9.9

40.0/

24.9

16.0/

9.9

16.0/

9.9

20.0/

12.4

20.0/

12.4

50.0/

31.1

40.0/

24.9

40.0/

24.9

30.0/

18.6

30.0/

18.6

60.0/

37.3

50.0/

31.1

50.0/

31.1

40.0/

24.9

40.0/

24.9

60.0/

37.3

60.0/

37.3

50.0/

31.1

50.0/

31.1

70.0/

43.5

70.0/

43.5

60.0/

37.3

60.0/

37.3

80.0/

49.7

80.0/

49.7

The Agency’s characterization testing (Tables 5 and 6) revealed that PAEB system

performance generally degraded in dark conditions compared to daylight conditions.

Additionally, certain test conditions, such as S1d and S1e, were particularly challenging in dark

conditions, especially when the vehicle’s lower beams were used. However, a few vehicles were

able to repeatedly avoid contact with the pedestrian mannequins at speeds up to 60 kph (37.3

mph) for certain test conditions when the vehicles’ lower beams provided the only source of

light.

NHTSA’s findings for PAEB system performance during testing align generally well

with those from IIHS’ recent system effectiveness study for 2017-2020 model year vehicles.

IIHS found that although PAEB systems were associated with a 32 percent reduction in

pedestrian crashes occurring during daylight, and a 33 percent reduction in pedestrian crashes for

areas with artificial lighting during dawn, dusk, or at night, there was no evidence that PAEB

systems were effective at nighttime without street lighting.

116

Based on the results of the PAEB characterization study and IIHS’ findings in its recent

study, NHTSA is proposing to perform the proposed test conditions (S1 a-e and S4 a-c) under

daylight conditions and under dark conditions with the vehicle’s lower beams. NHTSA notes

that Euro NCAP conducts PAEB testing that is similar to the Agency’s S4c test condition under

dark conditions with vehicles’ upper beams in use. Because the Agency cannot be assured that a

vehicle’s upper beams are in use during nighttime (i.e., dark lighting conditions) real-world

driving, NHTSA is proposing only to perform nighttime PAEB assessments using vehicles’

lower beams for all test conditions included in NCAP at this time. However, if the SV is

equipped with advanced lighting systems such as semiautomatic headlamp beam switching

and/or adaptive driving beam head lighting system, they shall be enabled to automatically engage

during the nighttime PAEB assessment. The Agency believes this approach covers the two

extreme light conditions and as such, information regarding performance with the upper beams

or under infrastructure lighting can be reasonably inferred.

The Agency recognizes that Euro NCAP performs testing similar to S1a and S1c at

speeds of 10 kph (6.2 mph) to 60 kph (37.3 mph) in dark conditions with the SV lower beams in

use; however, overhead streetlights are also used in these tests to provide additional light source.

To study potential performance differences attributable to the use of overhead lights during dark

conditions, NHTSA performed additional testing for PAEB scenarios S1 b, d, and e and S4 a and

c for a subset of test speeds, 16 kph (9.9 mph) and 40 kph (24.9 mph), for two of the MY 2020

vehicles used in its initial characterization study. This study was performed using the vehicles’

lower beams under dark conditions with overhead lights. For this limited testing, the Agency

116

Cicchino, J. B (2022, February), Effects of automatic emergency braking systems on pedestrian crash risk,

Insurance Institute for Highway Safety, https://www.iihs.org/api/datastoredocument/bibliography/2243.

observed slightly better PAEB performance in dark lighting conditions with overhead lights than

in dark lighting conditions without overhead lights.

NHTSA believes that testing with the vehicles’ lower beams in dark conditions without

overhead lights is appropriate, particularly at higher test speeds, as it would assure system

performance for real-world situations where visibility is the most limited. Furthermore, as

mentioned previously, dark lighting conditions with no overhead lights represented 36 percent of

pedestrian fatalities and dark lighting conditions with overhead lights represented 39 percent of

pedestrian fatalities in Volpe’s 2011-2015 FARS data set. Additionally, PAEB systems that

meet the performance test specifications under dark lighting conditions with no overhead lights

are likely to meet the performance specifications under dark lighting conditions with overhead

lights. Thus, the Agency believes assessment of PAEB systems under dark conditions with no

overhead lights and with the vehicle’s lower beams will encourage vehicle manufacturers to

make design improvements to address a significant portion of crashes that currently result in

pedestrian fatalities.

For the PAEB performance criteria, NHTSA is proposing that a vehicle must achieve

complete crash avoidance (i.e., have no contact with the pedestrian mannequin) in order to pass a

test trial conducted at each specified test speed (i.e., 10, 20, 30, 40, 50, and 60 kph (6.2, 12.4,

18.6, 24.9, 31.1, and 37.3 mph)) for each test condition (S1a, b, c, d, and e and S4a, b, and c).

NHTSA believes that this approach, used in conjunction with an incremental increase in SV

speed, should limit damage to the pedestrian mannequin and/or the SV during testing.

Along these lines, NHTSA is proposing a fourth change to the 2019 PAEB test procedure

regarding the number of test trials conducted for each combination of test condition and test

speed. The 2019 PAEB test procedure specifies seven test trials be conducted for each test speed

under each test condition. The Agency is proposing, however, to not require that more than one

test be conducted per test speed and test condition combination if certain criteria are met, and is

proposing that the pass rate for a given test speed will be dependent on whether additional test

trials are required to be performed.

117

For a given test condition, the test sequence is initiated at the 10 kph (6.2 mph) minimum

speed. To achieve a pass result, the test must be valid (i.e., all test specification and tolerances

satisfied), and the SV must not contact the pedestrian mannequin. If the SV does not contact the

pedestrian mannequin during the first valid test, the test speed is incrementally increased by 10

kph (6.2 mph), and the next test in the sequence is performed. Unless the SV contacts the

pedestrian mannequin, this iterative process continues until a maximum test speed of 60 kph

(37.3 mph) is evaluated. If the SV contacts the pedestrian mannequin, and the relative

longitudinal velocity between the SV and pedestrian mannequin is less than or equal to 50

percent of the initial speed of the SV, the Agency will perform four additional (repeated) test

trials at the same speed for which the impact occurred. The vehicle must not contact the

pedestrian mannequin for at least three out of the five test trials performed at that same speed to

pass that specific combination of test condition and test speed.

118

If the SV contacts the

pedestrian mannequin during a valid test of a test condition (whether it be the first test performed

for a particular test speed or a subsequent test trial at that same speed), and the relative impact

velocity exceeds 50 percent of the initial speed of the SV, no additional test trials will be

conducted at the given test speed and test condition and the SV is considered to have failed the

test condition at that specific test speed.

The Agency is pursuing an assessment approach for PAEB systems that differs from the

evaluation criteria proposed for the other four proposed ADAS technologies discussed earlier in

an attempt to reduce test burden, but still ensure that passing systems include robust designs that

117

This is a divergence from assessment of LKS, BSW, and BSI where a vehicle must meet performance

requirements for five out of seven valid test trials for a particular test condition to pass that test condition.

118

The Agency notes that a similar pass/fail criterion (i.e., a vehicle must meet performance requirements for three

out of five trials for a particular test condition to pass the test condition) is included in its LDW test procedure, as

referenced earlier.

will afford an enhanced level of safety. NHTSA recognizes that it is proposing a large number

of PAEB test conditions for inclusion in NCAP – eight total. The Agency also acknowledges

that these test conditions must be repeated for multiple test speeds and lighting conditions, which

inherently imposes additional test burden. Therefore, the Agency believes that it is reasonable to

reduce the number of test trials that must be conducted at a given test speed for a particular test

condition since the SV’s PAEB system will also be assessed at subsequent test speeds, which

would help system robustness. This would further be supported by the Agency’s proposal to

require that five test trials be performed in instances where the SV is unable to meet the no

contact performance requirement in the initial valid trial for that combination of test condition

and speed.

Although NHTSA believes that the assessment approach for PAEB systems proposed

herein is the most reasonable one, the Agency is requesting comment on whether it should

instead pursue an alternative approach, such as conducting seven trials for each test condition

and speed combination, and requiring that five of the seven trials meet the no contact

performance criterion. Again, this latter approach would be similar to the one proposed for the

other ADAS technologies discussed earlier.

Previously, NHTSA noted that it did not conduct the S2 and S3 test scenarios as part of

the characterization study and is not proposing these test scenarios for inclusion in this proposal.

The Agency agrees with the comments mentioned previously that the majority of vehicles in the

U.S. fleet are not currently equipped with sensing systems capable of detecting pedestrians while

a vehicle is turning, as they do not have the necessary FOV. The American Automobile

Association (AAA)

119

recently conducted PAEB tests, including an S2 scenario where the

vehicle is turning right with an adult pedestrian crossing. The PAEB systems in four model year

119

American Automobile Association (2019, October), Automatic emergency braking with pedestrian detection,

https://www.aaa.com/AAA/common/aar/files/Research-Report-Pedestrian-Detection.pdf.

2019 vehicles that were tested did not react to the test targets during a testing scenario that is

similar to NHTSA’s S2 scenario described above, resulting in all test vehicles colliding with the

pedestrian target. These systems performed better in a scenario that was similar to NHTSA’s S1;

however, the vehicles avoided a collision with the pedestrian target 40 percent of the time at a

32.2 kph (20 mph) test speed and nearly all the time at a 48.3 kph (30 mph) test speed.

Furthermore, in its recent study on PAEB system effectiveness, IIHS found that while AEB with

pedestrian detection was associated with significant reductions in pedestrian crash risk (~27

percent) and pedestrian injury crash risk (~30 percent), there was no evidence to suggest that

existing systems were effective while the PAEB-equipped vehicle was turning.

120

Considering

these findings, NHTSA believes that it is more beneficial at this time to focus our efforts on

performing PAEB testing at higher speeds and with various lighting conditions using the

proposed S1 and S4 test scenarios.

In the context of the NCAP PAEB testing program, NHTSA is seeking comment on the

following:

(23) Is the proposed test speed range, 10 kph (6.2 mph) to 60 kph (37.3 mph), to be assessed

in 10 kph (6.2 mph) increments, most appropriate for PAEB test scenarios S1 and S4?

Why or why not?

(24) The Agency has proposed to include Scenarios S1 a-e and S4 a-c in its NCAP

assessment. Is it necessary for the Agency to perform all test scenarios and test

conditions proposed in this RFC notice to address the safety problem adequately, or

could NCAP test only certain scenarios or conditions to minimize test burden but still

address an adequate proportion of the safety problem? Why or why not? If it is not

necessary for the Agency to perform all test scenarios or test conditions, which

120

Cicchino, J. B (2022, February), Effects of automatic emergency braking systems on pedestrian crash risk,

Insurance Institute for Highway Safety, https://www.iihs.org/api/datastoredocument/bibliography/2243.

scenarios/conditions should be assessed? Although they are not currently proposed for

inclusion, should the Agency also adopt the false positive test conditions, S1f and S1g?

Why or why not?

(25) Given that a large portion of pedestrian fatalities and injuries occur under dark lighting

conditions, the Agency has proposed to perform testing for the included test conditions

(i.e., S1 a-e and S4 a-c) under dark lighting conditions (i.e., nighttime) in addition to

daylight test conditions for test speed range 10 kph (6.2 mph) to 60 kph (37.3 mph).

NHTSA proposes that a vehicle’s lower beams would provide the source of light during

the nighttime assessments. However, if the SV is equipped with advanced lighting

systems such as semiautomatic headlamp beam switching and/or adaptive driving beam

head lighting system, they shall be enabled to automatically engage during the

nighttime PAEB assessment. Is this testing approach appropriate? Why or why not?

Should the Agency conduct PAEB evaluation tests with only the vehicle’s lower beams

and disable or not use any other advanced lighting systems?

(26) Should the Agency consider performing PAEB testing under dark conditions with a

vehicle’s upper beams as a light source? If yes, should this lighting condition be

assessed in addition to the proposed dark test condition, which would utilize only a

vehicle’s lower beams along with any advanced lighting system enabled to

automatically engage, or in lieu of the proposed dark testing condition? Should the

Agency also evaluate PAEB performance in dark lighting conditions with overhead

lights? Why or why not? What test scenarios, conditions, and speed(s) are appropriate

for nighttime (i.e., dark lighting conditions) testing in NCAP, and why?

(27) To reduce test burden in NCAP, the Agency proposed to perform one test per test speed

until contact occurs, or until the vehicle’s relative impact velocity exceeds 50 percent of

the initial speed of the subject vehicle for the given test condition. If contact occurs and

if the vehicle’s relative impact velocity is less than or equal to 50 percent of the initial

SV speed for the given combination of test speed and test condition, an additional four

test trials will be conducted at the given test speed and test condition, and the SV must

meet the passing performance criterion (i.e., no contact) for at least three out of those

five test trials in order to be assessed at the next incremental test speed. Is this an

appropriate approach to assess PAEB system performance in NCAP, or should a certain

number of test trials be required for each assessed test speed? Why or why not? If a

certain number of repeat tests is more appropriate, how many test trials should be

conducted, and why?

(28) Is a performance criterion of “no contact” appropriate for the proposed PAEB test

conditions? Why or why not? Alternatively, should the Agency require minimum

speed reductions or specify a maximum allowable SV-to-mannequin impact speed for

any or all of the proposed test conditions (i.e., test scenario and test speed

combination)? If yes, why, and for which test conditions? For those test conditions,

what speed reductions would be appropriate? Alternatively, what maximum allowable

impact speed would be appropriate?

(29) If the SV contacts the pedestrian mannequin during the initial trial for a given test

condition and test speed combination, NHTSA proposes to conduct additional test trials

only if the relative impact velocity observed during that trial is less than or equal to 50

percent of the initial speed of the SV. For a test speed of 60 kph (37.3 mph), this

maximum relative impact velocity is nominally 30 kph (18.6 mph), and for a test speed

of 10 kph (6.2 mph), the maximum relative impact velocity is nominally 5 kph (3.1

mph). Is this an appropriate limit on the maximum relative impact velocity for the

proposed range of test speeds? If not, why? Note that the tests in Global Technical

Regulation (GTR) No. 9 for pedestrian crashworthiness protection simulates a

pedestrian impact at 40 kph (24.9 mph).

(30) For each lighting condition, the Agency is proposing 6 test speeds (i.e., those performed

from 10 to 60 kph (6.2 to 37.3 mph) in increments of 10 kph (6.2 mph)) for each of the

8 proposed test conditions (S1a, b, c, d, and e and S4a, b, and c). This results in a total

of 48 unique combinations of test conditions and test speeds to be evaluated per

lighting condition, or 96 total combinations for both light conditions. The Agency

mentions later, in the ADAS Ratings System section, that it plans to use check marks,

as is done currently, to give credit to vehicles that (1) are equipped with the

recommended ADAS technologies, and (2) pass the applicable system performance test

requirements for each ADAS technology included in NCAP until it issues (1) a final

decision notice announcing the new ADAS rating system and (2) a final rule to amend

the safety rating section of the vehicle window sticker (Monroney label). For the

purposes of providing credit for a technology using check marks, what is an appropriate

minimum overall pass rate for PAEB performance evaluation? For example, should a

vehicle be said to meet the PAEB performance requirements if it passes two-thirds of

the 96 unique combinations of test conditions and test speeds for the two lighting

conditions (i.e., passes 64 unique combinations of test conditions and test speeds)?

(31) Given previous support from commenters to include S2 and S3 scenarios in the

program at some point in the future and the results of AAA’s testing for one of the

turning conditions, NHTSA seeks comment on an appropriate timeframe for including

S2 and S3 scenarios into the Agency’s NCAP. Also, NHTSA requests from vehicle

manufacturers information on any currently available models designed to address, and

ideally achieve crash avoidance during conduct of, the S2 and S3 scenarios to support

Agency evaluation for a future program upgrade.

(32) Should the Agency adopt the articulated mannequins into the PAEB test procedure as

proposed? Why or why not?

(33) In addition to tests performed under daylight conditions, the Agency is proposing to

evaluate the performance of PAEB systems during nighttime conditions where a large

percentage of real-world pedestrian fatalities occur. Are there other technologies and

information available to the public that the Agency can evaluate under nighttime

conditions?

(34) Are there other safety areas that NHTSA should consider as part of this or a future

upgrade for pedestrian protection?

(35) Are there any aspects of NCAP’s proposed PAEB test procedure that need further

refinement or clarification before adoption? If so, what additional refinement or

clarification is necessary, and why?

In addition to the fleet characterization research conducted for this upgrade of NCAP, the

Agency is conducting additional research that may be used to support future program

enhancements. One such research project is designed to address injuries and fatalities for other

vulnerable road users, specifically cyclists.

121

While some PAEB systems may be capable of

detecting cyclists and activating to avoid a crash, NHTSA’s current PAEB test procedure does

not include a specific cyclist component. However, since the number of cyclists killed on U.S.

roads continues to rise,

122

the Agency plans to perform research to determine the viability of

Euro NCAP’s AEB cyclist tests. NHTSA will then compare test data with preliminary crash

populations to assess the adequacy of the test procedure for the U.S. vehicle fleet and roadway

system. The Euro NCAP test includes four test scenarios: one in which the cyclist crosses in

front of the vehicle from the near-side; one in which the cyclist crosses in front of the vehicle

121

NHTSA notes that this research will also include motorcycles.

122

National Center for Statistics and Analysis (2019, June), Bicyclists and other cyclists: 2017 data (Traffic Safety

Facts. Report No. DOT HS 812 765), Washington, DC: National Highway Traffic Safety Administration.

from the near-side from behind an obstruction; one in which the cyclist crosses in front of the

vehicle from the far-side; and the other in which the cyclist travels in the same direction as the

vehicle. The latter test scenario is repeated for both 25 percent and 50 percent overlaps, while

the first three scenarios are conducted at 50 percent overlap (i.e., the vehicle strikes the bicyclist

at 50 percent of the vehicle’s width). In all tests, a cyclist target comprised of an articulating

dummy, which replicates the pedaling action of a cyclist, is seated on a bicycle mounted on a

moving platform.

NHTSA believes that detecting cyclists is technically more challenging for vehicle AEB

systems than detecting pedestrians since cyclists often move at higher speeds. Vehicles must not

only be equipped with sensors that have wider fields of view (similar to that required for the

turning PAEB test scenarios), but must also process information more quickly as to whether to

alert the driver and/or automatically brake.

In the context of this additional research testing, NHTSA requests comment on the

following:

(36) Considering not only the increasing number of cyclists killed on U.S. roads but also the

limitations of current AEB systems in detecting cyclists, the Agency seeks comment on

the appropriate timeframe for adding a cyclist component to NCAP and requests from

vehicle manufacturers information on any currently available models that have the

capability to validate the cyclist target and test procedures used by Euro NCAP to

support evaluation for a future NCAP program upgrade.

(37) In addition to the test procedures used by Euro NCAP, are there others that NHTSA

should consider to address the cyclist crash population in the U.S. and effectiveness of

systems?

D. Updating Forward Collision Prevention Technologies

As previously mentioned, NHTSA will retain the currently available ADAS technologies

(forward collision warning, crash imminent braking and dynamic brake support) designed to

address forward collisions (rear-end crashes) in NCAP’s crash avoidance program. As discussed

in NHTSA’s March 2019 study, these technologies have the potential to prevent or mitigate eight

rear-end pre-crash scenarios, which represented approximately 1.70 million crashes annually, on

average, or 29.4 percent of all crashes that occurred on U.S. roadways. As shown in Table A-1,

these crashes resulted in 1,275 fatalities, on average, and 883,386 MAIS 1-5 injuries annually,

which represented 3.8 percent of all fatalities and 31.5 percent of all injuries, respectively.

123

FCW technology evaluations were introduced into NCAP starting with model year 2011

vehicles,

124

while CIB and DBS systems (referred to collectively as Automatic Emergency

Braking (AEB)) were added to the program starting with model year 2018 vehicles.

125

These

technologies are not being offered as standard equipment on all passenger vehicles, so it remains

important for NCAP to recommend the technologies and inform shoppers which vehicles have

the technologies. Further, NHTSA observed performance test failures for each of these

technologies during NCAP’s model year 2019 vehicle performance verification testing;

126

thus,

NCAP should continue to inform shoppers as to which systems perform to NHTSA’s

benchmark. Nonetheless, as will be discussed in the next few sections, NHTSA believes there

are opportunities for updating the current NCAP performance requirements for these three

technologies.

1. Forward Collision Warning (FCW)

An FCW system is an ADAS technology that monitors a vehicle’s speed, the speed of the

vehicle in front of it, and the distance between the two vehicles. If the FCW system determines

123

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653). Washington, DC: National Highway Traffic Safety Administration.

124

73 FR 40016 (July 11, 2008).

125

80 FR 68618 (Nov. 5, 2015).

126

https://www.regulations.gov, Docket Nos. NHTSA-2010-0093 and NHTSA-2015-0006. (Only one test failure

was observed for FCW.)

that the distance from the driver’s vehicle to the vehicle in front of it is too short, and the closing

velocity between the two vehicles is too high, the system warns the driver of an impending rear-

end collision.

Typically, FCW systems are comprised of two components: a sensing system, which can

detect a vehicle in front of the driver’s vehicle; and a warning system, which alerts the driver to a

potential crash threat. The sensing portion of the system may consist of forward-looking radar,

lidar, camera systems, or a combination of these. The warning system may provide drivers with

a visual display, such as a light on the dash, an audible signal (e.g., buzzer or chime), and/or a

haptic signal that provides tactile feedback to the driver (e.g., rapid vibrations of the seat pan or

steering wheel) to alert the driver of an impending crash so that they may manually intervene

(e.g., apply the vehicle’s brakes or make an evasive steering maneuver) to avoid or mitigate the

crash.

Currently, NCAP’s FCW test procedure

127

consists of three scenarios that simulate the

most frequent types of rear-end crashes. These include: lead vehicle stopped (LVS), lead vehicle

decelerating (LVD), and lead vehicle moving (LVM) scenarios. In each scenario, the vehicle

being evaluated is the SV, and the vehicle positioned directly in front of the SV, a production

mid-size passenger car, is the POV. The time-to-collision (TTC) criteria prescribed for each

scenario represent the time needed for a driver to perceive an impending rear-end crash, decide

the corrective action, and respond with the appropriate mitigating action. The TTC for each

scenario is calculated by considering the speed of the SV relative to the POV at the time of the

FCW alert. If the FCW system fails to provide an alert within the required time during testing,

the professional test driver brakes or steers away to avoid a collision. A short description of each

test scenario and the requirements for a passing result based on TTC is provided below:

127

National Highway Traffic Safety Administration. (2013, February). Forward collision warning system

confirmation test. https//www.regulations.gov. Docket No. NHTSA-2006-26555-0134.

 LVS – The SV encounters a stopped POV on a straight road. The SV is moving at 72.4

kph (45 mph), and the POV is stationary. To pass this test, the SV must issue an FCW

alert when the TTC is at least 2.1 s.

 LVD – The SV encounters a POV slowing with constant deceleration directly in front of

it on a straight road. The SV and POV are both driven at 72.4 kph (45 mph) with an

initial headway of 30.0 m (98.4 ft.). The POV then decelerates, braking at a constant

deceleration of 0.3g in front of the SV. To pass this test, the SV must issue an FCW alert

when the TTC is at least 2.4 s.

 LVM – The SV encounters a slower-moving POV directly in front of it on a straight

road. The SV and POV are driven at constant speeds of 72.4 kph (45 mph) and 32.2 kph

(20 mph), respectively. To pass this test, the SV must issue an FCW alert when the TTC

is at least 2.0 s.

Each scenario is conducted up to seven times. To pass the NCAP system performance

criteria, the SV must pass at least five out of seven trials

128

for each of the three test scenarios.

NCAP’s FCW test scenarios are directly related to real-world crash data. From its

analysis of 2011 to 2015 FARS and GES data, the Agency found that crashes analogous to the

LVS test scenario, where a struck vehicle was stopped at the time of impact, occurred in 65

percent of the rear-end crashes studied.

129

The LVD scenario, in which the struck vehicle was

decelerating at the time of impact, occurred in 22 percent of the rear-end crashes, and the LVM

scenario, in which the struck vehicle was moving at a constant, but slower, speed compared to

128

As noted in the Agency’s 2015 AEB final decision notice (80 FR 68618 (Nov. 5, 2015)), the Agency believes

passing five out of seven tests successfully discriminates between functional systems versus non-functional systems.

To date, the Agency allows two failures out of seven attempts to afford some flexibility in including emerging

technologies into the NCAP program. Furthermore, NHTSA test laboratories have experienced unpredictable

vehicle responses due to the vehicle algorithm designs. Test laboratories have observed systems that improve their

performance with use, systems degrading and shutting down when they do not see other vehicles, and systems

failing to re-activate if the vehicle is not cycled through an ignition cycle.

129

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

the striking vehicle at impact, occurred in 10 percent of the rear-end crashes. Collectively, these

test scenarios represented 97 percent of rear-end crashes. With respect to test speed, in its

independent review of the 2011-2015 FARS and GES data sets, Volpe concluded that 28 percent

of fatal rear-end crashes and 63 percent of all rear-end crashes occurred on roadways with posted

speed limits of 72.4 kph (45 mph) or less.

Currently, NHTSA gives credit on its website by assigning a check mark to vehicles

equipped with FCW systems that send visual, audible, and/or haptic alerts and meet the TTC

requirements. However, the Agency’s research has shown that presenting drivers with an

audible warning in medium or high urgency situations significantly reduced crash severity

relative to visual and tactile (or haptic) warnings, which did not differ.

130

This being said, in a

large-scale field test of FCW and LDW systems on model year 2013 Chevrolet and Cadillac

vehicles, the University of Michigan Transportation Research Institute (UMTRI) and GM found

that GM’s Safety Alert Seat, which provides haptic seat vibration pulses, increased driver

acceptance of both FCW and LDW systems compared to audible alerts.

131

The study concluded

that the FCW system was turned off 6 percent of the time when the Safety Alert Seat was

selected (rather than audible alerts), whereas it was turned off 17 percent of the time when only

audible alerts were available. In light of these findings, the Agency seeks comment on whether

to give credit to vehicles equipped with FCW systems that only provide a passing audible alert,

or whether it should also give credit to those systems that only provide passing haptic alerts.

132

If the Agency elects to give credit to vehicles with haptic alerts, are there certain haptic alert

130

Lerner, N., Robinson, E., Singer, J., Jenness, J., Huey, R., Baldwin, C., & Fitch, G. (2014, September), Human

factors for connected vehicles: Effective warning interface research findings (Report No. DOT HS 812 068),

Washington, DC: National Highway Traffic Safety Administration.

131

Flannagan, C., LeBlanc, D., Bogard, S., Nobukawa, K., Narayanaswamy, P., Leslie, A., Kiefer, R., Marchione,

M., Beck, C., and Lobes, K. (2016, February), Large-scale field test of forward collision alert and lane departure

warning systems (Report No. DOT HS 812 247), Washington, DC: National Highway Traffic Safety

Administration.

132

The Agency would give credit to FCW systems that have both passing audible and haptic alerts if both alert types

were available. However, if a vehicle with such a system provided only a passing haptic alert and the Agency

decided only to give credit to systems that provided passing audible alerts, then the vehicle would not receive credit

as having met the Agency’s FCW test requirements.

types that should be excluded from consideration (e.g., because they may be such a nuisance to

drivers that they may be more likely to disable the system)? NHTSA also seeks comment on

whether it should no longer give credit to FCW-equipped vehicles that offer only visual FCW

alerts.

NCAP’s current FCW test procedure states that if an FCW system provides a warning

timing adjustment setting for the driver, at least one timing setting must meet the TTC warning

criteria specified in the procedure. Therefore, if a vehicle is equipped with a warning timing

adjustment, only the most conservative (i.e., earliest) warning setting is tested. Selecting the

most conservative setting is beneficial for track testing where the driver of the SV must steer

and/or brake to avoid a crash with the POV after the FCW alert is issued. However, the Agency

is concerned that many consumers may not adjust the warning timing setting for FCW alerts.

Furthermore, consumers that choose to adjust the alert timing may be unlikely to select the

earliest setting, as this setting is most likely to result in false positive alerts (i.e., nuisance alerts)

during real-world operation.

133

The Agency also recognizes that the earliest FCW setting can be

used to pass the NCAP test—essentially allowing a vehicle to get NCAP credit even though it

may not otherwise earn credit if the later warning settings are tested. Therefore, by testing the

earliest timing adjustment setting, the Agency’s FCW performance assessment may not be

indicative of many drivers’ real-world experiences.

This concern was previously addressed in NHTSA’s 2015 AEB final decision notice, but

the Agency has not since made updates to its FCW test procedure.

134

In that notice, the Agency

stated that because NCAP is a consumer information program, it should test vehicles as

delivered, using the factory default FCW warning adjustment setting for FCW and AEB testing,

133

Nodine, E., Fisher, D., Golembiewski, G., Armstrong, C., Lam, A., Jeffers, M.A., Najm, W., Miller, S., Jackson,

S., and Kehoe, N. (2019, May), Indicators of driver adaptation to forward collision warnings: A naturalistic driving

evaluation (Report No. DOT HS 812 611), Washington, DC: National Highway Traffic Safety Administration.

134

80 FR 68614 (Nov. 5, 2015).

including PAEB. Although the Agency believes there is still merit to testing the default setting,

NHTSA tentatively believes testing the middle alert setting may be more appropriate. Selection

of the middle or next latest alert setting for testing would harmonize with Euro NCAP’s AEB

Car-to-Car systems test protocol, thus potentially driving costs down for manufacturers and

attempting to ensure that consumers in both the U.S. and European markets benefit from similar

FCW system settings.

135

Harmonization was a common theme among commenters responding

to NCAP’s December 2015 notice, with most vehicle manufacturers, suppliers, and other

industry groups requesting that NHTSA harmonize test procedures, test targets, and test

requirements with other NCAPs around the world, particularly Euro NCAP. As mentioned

earlier, the Bipartisan Infrastructure Law also required that NHTSA consider harmonization with

third-party safety rating programs when possible. In light of these considerations, the Agency is

proposing that it is most appropriate to test the middle (or next latest) FCW system setting in lieu

of the default setting when performing FCW, CIB, DBS, and PAEB NCAP tests on vehicles that

offer multiple FCW timing adjustment settings.

FCW systems have been recognized as the first generation of ADAS technologies

designed to help drivers avoid an impending rear-end collision. In 2008, when NHTSA decided

to include ADAS in the NCAP program, FCW was selected because the Agency believed (1) this

technology addressed a major crash problem; (2) system designs existed that could mitigate this

safety problem; (3) safety benefit projections were assessed; and (4) performance tests and

procedures were available to ensure an acceptable performance level.

136

At the time, the Agency

estimated that FCW systems were 15 percent effective in preventing rear-end crashes. More

recently, in a 2017 study, IIHS

137

found that FCW systems may be more effective than NHTSA’s

135

European New Car Assessment Programme (Euro NCAP) (2019, July), Test Protocol – AEB Car-to-Car systems,

Version 3.0.2. See section 7.4.1.1.

136

73 FR 40033 (July 11, 2008).

137

Cicchino, J. B. (2017, February), Effectiveness of forward collision warning and autonomous emergency braking

systems in reducing front-to-rear crash rates, Accident Analysis and Prevention, 2017 Feb;99(Pt A):142-152.

https://doi.org/10.1016/j.aap.2016.11.009.

initial estimates. IIHS found that FCW systems reduced rear-end crashes by 27 percent.

Moreover, consumers have shown favorable acceptance of these systems. For instance, in a

2019 survey of more than 57,000 Consumer Reports subscribers, 69 percent of vehicle owners

reported that they were satisfied with their vehicle’s FCW technology, 38 percent of vehicle

owners said that it had helped them avoid a crash, and 54 percent of them remarked that they

trust the system to work every time.

138

As consumer acceptance has been positive, and system

performance has improved over the years, fitment rates have also increased. As mentioned

previously, less than 0.2 percent of model year 2011 vehicles were equipped with FCW systems

compared to 38.3 percent of model year 2018 vehicles.

One limitation of FCW systems is that they are designed to warn the driver, but not to

provide significant automatic braking of the vehicle (some FCW systems use haptic brake pulses

to alert the driver of a crash-imminent driving situation, but they are not intended to effectively

slow the vehicle). Since the introduction of FCW systems into NCAP, active safety systems,

such as those with automatic braking capability (i.e., AEB), have entered the marketplace. In a

recent study sponsored by GM

139

to evaluate the real-world effectiveness of ADAS technologies

(including FCW and AEB) on 3.8 million model year 2013-2017 GM vehicles, UMTRI found

that, for frontal collisions, camera-based FCW systems produced an estimated 21 percent

reduction in rear-end striking crashes, while the AEB systems studied (which included a

combination of camera-only, radar-only, and fused camera-radar systems) produced an estimated

46 percent reduction in the same crash type.

140

Similarly, in a 2017 study, IIHS found that

138

Consumer Reports (2019, August 5), Guide to forward collision warning: How FCW helps drivers avoid

accidents, https://www.consumerreports.org/car-safety/forward-collision-warning-guide/.

139

Leslie, A. J., Kiefer, R. J., Meitzner, M. R., & Flannagan, C. A. (2019), Analysis of the field effectiveness of

General Motors production active safety and advanced headlighting systems, The University of Michigan

Transportation Research Institute and General Motors LLC. UMTRI-2019-6.

140

The Agency notes that the FCW effectiveness rate (21%) observed by UMTRI is similar to that observed by IIHS

in its 2019 study (27%). Differences in data samples and vehicle selection may contribute to the specific numerical

differences. Regardless, the AEB effectiveness rate observed by UMTRI (46%) was significantly higher than the

corresponding FCW effectiveness rate observed in either the IIHS or UMTRI study.

vehicles equipped with FCW and AEB showed a 50 percent reduction for the same crash type.

141

NHTSA is drawing from these research studies, generally, since each has limitations and

deviations from how NHTSA might evaluate fleet-wide

142

system effectiveness.

From a functional perspective, research suggests that active braking systems, such as AEB,

provide greater safety benefits than corresponding warning systems, such as FCW. However,

NHTSA has found that current AEB systems often integrate the functionalities of FCW and AEB

into one frontal crash prevention system to deliver improved real-world safety performance and

high consumer acceptance. Consequently, the Agency believes that this system integration may

have implications for NCAP FCW testing because current NCAP FCW requirements were

developed at a time when FCW and AEB functionalities were not always linked. As will be

detailed later in this notice, NHTSA believes that FCW could now be considered a component of

AEB and PAEB such that FCW operation could be evaluated using NCAP’s AEB and PAEB

tests.

2. Automatic Emergency Braking (AEB)

To address the rear-end crash problem further, in November 2015, NHTSA published a

final decision notice announcing the addition of two AEB technologies, CIB and DBS, into

NCAP effective with model year 2018 vehicles.

143

Unlike FCW systems, AEB systems (i.e., CIB and DBS), are designed to help drivers

actively avoid or mitigate the severity of rear-end crashes. CIB systems provide automatic

braking when forward-looking sensors indicate that a crash is imminent and the driver has not

braked, whereas DBS systems provide supplemental braking when sensors determine that driver-

applied braking is insufficient to avoid an imminent crash.

141

Low-speed AEB showed a 43% reduction.

142

The UMTRI study was limited to GM vehicles.

143

80 FR 68604 (Nov. 5, 2015). CIB and DBS together are considered Automatic Emergency Braking (AEB).

In Consumer Reports’ 2019 subscriber survey, 81 percent of vehicle owners reported that

they were satisfied with AEB technology, 54 percent said that it had helped them avoid a crash,

and 61 percent stated that they trusted the system to work every time.

144

Furthermore, IIHS

found in its 2017 study that rear-end collisions decreased by 50 percent for vehicles equipped

with AEB and FCW.

145

Similarly, as mentioned earlier, UMTRI

146

found that AEB systems

produced an estimated 46 percent reduction in applicable rear-end crashes when combined with a

forward collision alert, which alone showed only a 21 percent reduction.

147

A recent IIHS study

148

of 2009-2016 crash data from 23 States suggested that the

increasing effectiveness of AEB technology in certain crash situations is changing the rear-end

crash problem. The Institute’s analysis provided insight into the performance of current AEB

systems and future opportunities for improvement. The study identified the types of rear-end

crashes in which striking vehicles equipped with AEB were over-represented compared to those

without AEB.

149

For instance, IIHS found that striking vehicles involved in the following rear-

end crashes were more likely to have AEB: (1) where the striking vehicle was turning relative to

when it was moving straight; (2) when the struck vehicle was turning or changing lanes relative

to when it was slowing or stopped; (3) when the struck vehicle was not a passenger vehicle or

was a special use vehicle relative to a passenger car; (4) on snowy or icy roads; or (5) on roads

with speed limits of 112.7 kph (70 mph) relative to those with 64.4 to 72.4 kph (40 to 45 mph)

144

Consumer Reports, (2019, August 5), Guide to automatic emergency braking: How AEB can put the brakes on

car collisions, https://www.consumerreports.org/car-safety/automatic-emergency-braking-guide/

145

Cicchino, J. B. (2017, February), Effectiveness of forward collision warning and autonomous emergency braking

systems in reducing front-to-rear crash rates, Accident Analysis and Prevention, 2017 Feb;99(Pt A):142-152,

https://doi.org/10.1016/j.aap.2016.11.009.

146

Leslie, A. J., Kiefer, R. J., Meitzner, M. R., & Flannagan, C. A. (2019, September), Analysis of the field

effectiveness of General Motors production active safety and advanced headlighting systems, The University of

Michigan Transportation Research Institute and General Motors LLC, UMTRI-2019-6.

147

The AEB systems studied by UMTRI consisted of camera-only, radar-only, and fused camera-radar AEB

systems, the latter two systems of which also included adaptive cruise control functionality.

148

Cicchino, J. B. & Zuby, D. S. (2019, August), Characteristics of rear-end crashes involving passenger vehicles

with automatic emergency braking, Traffic Injury Prevention, 2019, VOL. 20, NO. S1, S112–S118

https://doi.org/10.1080/15389588.2019.1576172.

149

In this instance, over-represented means a higher frequency as a percentage for AEB-equipped vehicles versus

non-AEB-equipped vehicles on a normalized basis.

speed limits. Overall, the study found that 25.3 percent of crashes where the striking vehicle was

equipped with AEB had at least one of these over-represented characteristics, compared with

15.9 percent of impacts by vehicles that were not equipped with AEB.

These results suggest that the tests used to evaluate the performance of AEB systems by

the Agency’s NCAP and other consumer information programs are influencing the development

of countermeasures capable of minimizing the crash problems that they were intended to address.

However, the results also imply that AEB systems have not yet provided their full crash

reduction potential. While they are effective at addressing the most common rear-end crashes,

they are less effective at addressing those crashes that are more atypical. IIHS found that in

2016, nearly 300,000 (15 percent) of the police reported two-vehicle rear-end crashes involved

one of the rear-end crashes mentioned above. The Institute suggested that vehicle manufacturers

would be encouraged to improve AEB system designs for situations where AEB was over-

represented if consumer programs incorporated tests that replicate these rear-end crash events,

such as an angled target vehicle that simulates a struck vehicle changing lanes. IIHS cautioned

(and NHTSA agrees) that new testing protocols should not drive performance degradation in

more typical crash situations, create unintended safety consequences, or adversely affect AEB

use due to nuisance activations.

While these recent studies suggest that AEB systems (i.e., CIB and DBS) have

collectively been effective in reducing rear-impact crashes, it is not clear how effective each of

these systems are as standalone systems, and whether their individual effectiveness may change

for certain crash scenarios, environmental conditions, or driver factors (e.g., poor judgement,

distraction, etc.). Furthermore, the Agency is not aware of any studies of current-generation

AEB systems that have determined the extent to which CIB and DBS individually contributes to

crash reduction.

Prior to considering adopting AEB into NCAP, NHTSA conducted a review of 2003-

2009 National Automotive Sampling System Crashworthiness Data System (NASS CDS) data to

define the target population for rear-end crashes.

150

At the time of the analysis, the Agency

concluded that CIB and DBS target crash populations were mutually exclusive. In other words,

they included crashes in which the driver either did not brake (CIB) or braked (DBS). The

analysis of the crash data showed that the driver braked in approximately half of the crashes and

did not brake in the other half. However, in its analysis of the 2011-2015 FARS and GES data

sets, Volpe found much more conservative brake rates. The organization found that the driver

braked in just 8 percent of rear-end crashes involving fatalities and 20 percent of those crashes

involving injuries. The study also showed that the driver made no attempt to avoid the crash

(e.g., no braking, steering, accelerating) for 56 percent of the crashes involving fatalities and for

21 percent of those involving injuries.

151

It is possible that the brake rate differed for the two

studies because of the target crash population refinements made for NHTSA’s original analysis

and because of difference in data collection methods between the crash databases. For instance,

high-speed crashes were excluded from NHTSA’s target crash population review because the

AEB systems tested at the time had limited speed reduction capabilities.

From the refined target crash population, NHTSA computed preliminary safety benefits

for both CIB and DBS from a limited number of CIB- and DBS-equipped vehicles subjected to

early versions of the Agency’s test procedures based upon speed reduction capabilities.

152

The

Agency recognized that CIB and DBS systems available at the time had limited capabilities and

could not address serious crashes where fatalities were likely to occur. Nevertheless, the Agency

tentatively found that if a CIB system alone was equipped on all light vehicles, it could

150

National Highway Traffic Safety Administration (2012, June), Forward-looking advanced braking technologies

research report, https://www.regulations.gov/document?D=NHTSA-2012-0057-0001.

151

The Agency notes that for the rear-end pre-crash scenario group, the driver avoidance maneuver was unknown in

25 percent and 54 percent of the FARS and GES crashes, respectively.

152

National Highway Traffic Safety Administration (2014, August), Automatic emergency braking system (AEB)

research report, https://www.regulations.gov/document?D=NHTSA-2012-0057-0037.

potentially prevent approximately 40,000 minor/moderate injuries (AIS 1 – 2), 640 serious-to-

critical injuries (AIS 3 – 5), and save approximately 40 lives, annually. If a DBS system alone

was equipped on all light vehicles, it could potentially prevent approximately 107,000

minor/moderate injuries (AIS 1 – 2), 2,100 serious-to-critical injuries (AIS 3 – 5), and save

approximately 25 lives, annually. These safety benefits from CIB and DBS were considered

incremental to the benefits stemming from an FCW alert.

153

NHTSA’s analysis showed there was merit to performing testing to assess vehicle

performance in situations where a driver either does not brake (CIB) or brakes (DBS). Volpe’s

recent analysis on braking behavior/rate further validates the need to assess CIB and DBS

separately. Considering this and the fact that NHTSA cannot currently differentiate the

individual effectiveness of CIB and DBS systems, NHTSA tentatively believes NCAP should

continue to assess CIB and DBS system performance individually. However, the Agency

acknowledges that, because it believes AEB systems have advanced significantly in recent years,

it is appropriate at this time to consider revising performance envelopes and dynamic scenarios

in NCAP to acknowledge and encourage such advances.

The following sections discuss in detail CIB and DBS systems, and more specifically,

NCAP’s current test procedures and a potential updated test program for modern AEB systems.

The Agency seeks comment on how NCAP can encourage the maximum safety benefits of AEB

and potentially reduce the number of tests conducted. Comments are also sought on future

suggestions for AEB beyond any near-term upgrade.

a. Dynamic Brake Support (DBS)

In response to an FCW alert or a driver noticing an imminent crash scenario, a driver may

initiate braking to avoid a rear-end crash. In situations where the driver’s braking is insufficient

153

FCW, CIB, and DBS combined on all light vehicles could potentially prevent approximately 200,000

minor/moderate injuries (AIS 1 – 2), 4,000 (AIS 3 – 5) serious injuries, and save approximately 100 lives annually.

to prevent a collision, DBS can automatically supplement the driver’s braking action to prevent

or mitigate the crash. Similar to FCW and CIB systems, DBS systems employ forward-looking

sensors such as radar, lidar, and/or vision-based sensors to detect vehicles in the path directly

ahead and monitor a vehicle’s operating conditions such as speed or brake application.

However, DBS systems can actively supplement braking to assist the driver whereas FCW

systems serve only to warn the driver of a potential crash threat, and CIB systems are activated

when a rear-end crash is imminent, but the driver has not manually applied the vehicle’s

brakes.

154

NCAP’s current DBS test procedure

155

consists of the same three rear-end crash scenarios

specified in the FCW system performance test procedure—LVS, LVD, and LVM, but most of

the test speed combinations specified in the DBS test procedure differ (the single exception is

that the FCW and DBS test procedures both use an LVM test performed with SV and POV

speeds of 72.4 and 32.2 kph (45 and 20 mph), respectively). In addition, the DBS performance

assessment includes a Steel Trench Plate (STP) false positive suppression test, which is

conducted at two test speeds. This fourth test scenario is used to evaluate the propensity of a

vehicle’s DBS system to activate inappropriately in a non-critical driving scenario that would not

present a safety risk to the vehicle’s occupants. For the first three test scenarios, where braking

is expected, the SV must provide enough supplemental braking to avoid contact with the POV to

pass a trial run. In the case of the DBS false positive test scenario, the performance criterion is

minimal to no activation for both test speeds.

156

154

DBS systems differ from traditional brake assist systems used with the vehicle’s foundation brakes. Whereas

both systems rely on brake pedal application rate to determine whether supplemental braking is required, DBS has a

lower activation threshold since it also uses information from the aforementioned sensors to verify that more braking

is needed.

155

National Highway Traffic Safety Administration (2015, October), Dynamic brake support performance

evaluation confirmation test for the New Car Assessment Program, http://www.regulations.gov, Docket No.

NHTSA-2015-0006-0026.

156

Minimal activation is defined as a peak SV deceleration attributable to DBS intervention that is less than or equal

to 1.25 times the average of the deceleration recorded for the vehicle’s foundation brake system alone during its

As in the FCW system performance tests, the vehicle that is subjected to the DBS test

scenarios is the SV. The FCW test procedure (which uses professional drivers for acceleration,

braking, and steering during test conduct) stipulates that a mid-size passenger car serve as the

POV during testing. The DBS test procedure (which relies solely on the use of a programmable

brake controller and the vehicle’s DBS system for braking), however, utilizes a surrogate (i.e.,

target vehicle) to limit the potential for damage to the SV and/or test equipment in the event of a

collision.

The target vehicle presently used as the POV by NCAP for the Agency’s DBS testing is

known as the Subject Surrogate Vehicle, or SSV. The SSV, developed by NHTSA for the

purpose of track testing, appears as a “real” vehicle to the camera, radar, and lidar sensors used

by existing AEB systems. The SSV system is comprised of (a) a shell,

157

which is a visually and

dimensionally accurate representation of a passenger car; (b) a slider and load frame assembly to

which the shell is attached, (c) a two-rail track on which the slider operates, (d) a road-based

lateral restraint track, and (e) a tow vehicle, which pulls the SSV and its peripherals down the

test track during trials where the POV (i.e., SSV) must be in motion. A brief discussion on the

use of the GVT, discussed earlier in the BSI section, as an alternative to the SSV for future DBS

and CIB testing, is included later in this notice.

158

A short description of each DBS system performance test scenario, and the requirements

for a passing result, is provided below:

approach to the steel trench plate. The 1.25 multiplier serves to provide some system flexibility, meaning a mild

DBS intervention is acceptable, but one where the vehicle thinks it must respond to the STP as if it was a real

vehicle is not.

157

The shell is constructed from lightweight composite materials with favorable strength-to-weight characteristics,

including carbon fiber, Kevlar®, phenolic, and Nomex honeycomb. It is also wrapped with a commercially available

vinyl material to simulate paint on the body panels, rear bumper, and a tinted glass rear window. A foam bumper

having a neoprene cover is attached to the rear of the SSV to reduce the peak forces realized immediately after an

impact from a test vehicle occurs.

158

If the Agency decides to assess FCW in separate tests to that for DBS and CIB, those FCW tests would also be

conducted using GVT.

 Lead Vehicle Stopped (LVS) – The SV encounters a stopped POV on a straight road.

The SV is moving at 40.2 kph (25 mph) and the POV is stationary. The SV throttle is

released within 500 ms after the SV issues an FCW alert, and the SV brake is applied at a

TTC of 1.1 s (i.e., at a nominal headway of 12.2 m (40 ft.)). To pass this test, the SV

must not contact the POV.

 Lead Vehicle Decelerating (LVD) – The SV encounters a POV slowing with constant

deceleration directly in front of it on a straight road. The SV and POV are both driven at

56.3 kph (35 mph) with an initial headway of 13.8 m (45.3 ft.). The POV brakes are then

applied at a constant deceleration of 0.3g in front of the SV. The SV throttle is released

within 500 ms after the SV issues an FCW alert, and the SV brakes are applied at a TTC

of 1.4 s (i.e., at a nominal headway of 9.6 m (31.5 ft.)). To pass this test, the SV must not

contact the POV.

 Lead Vehicle Moving (LVM) – The SV encounters a slower-moving POV directly in

front of it on a straight road. In the first test, the SV and POV are driven on a straight

road at a constant speed of 40.2 kph (25 mph) and 16.1 kph (10 mph), respectively. In

the second test, the SV and POV are driven at a constant speed of 72.4 kph (45 mph) and

32.2 kph (20 mph), respectively. In both tests, the SV throttle is released within 500 ms

after the SV issues an FCW alert, and the SV brakes are applied at a TTC of 1 s (i.e., at a

nominal headway of 6.7 m (22 ft.) in the first test, and 11.3 m (37 ft.) in the second test).

To pass these tests, the SV must not contact the POV.

 Steel Trench Plate (STP) test (to assess false positive suppression) – The SV is driven

over a 2.4 m x 3.7 m x 25.4 mm (8 ft. x 12 ft. x 1 in.) steel trench plate at 40.2 kph (25

mph) and 72.4 kph (45 mph). If no FCW alert is issued by a TTC of 2.1 s, the SV throttle

is released within 500 ms of a TTC of 2.1 s, and the SV brakes are applied at a TTC of

1.1 s (i.e.., at a nominal distance of 12.3 m (40 ft.) from the edge of the STP at 40.2 kph

100

(25 mph), or 22.3 m (73 ft.) at 72.4 kph (45 mph)). To pass this test, the performance

criteria is non-activation, as defined above.

To pass NCAP’s DBS system performance criteria, the SV must currently pass five out

of seven trials for each of the six test conditions.

As previously mentioned, NCAP’s LVS, LVM, and LVD test scenarios for its DBS

evaluations are similar to those for the FCW assessments and therefore correspond well with

real-world crash data and have similar target crash populations. NHTSA’s analysis of the 2011-

2015 rear-end crash data from FARS and GES showed target crash populations of 65 percent for

the LVS scenario, 22 percent for the LVD scenario, and 10 percent for the LVM scenario.

159

Furthermore, Volpe’s independent review of the 2011-2015 data sets showed that for rear-end

crashes that occurred on roadways with posted speeds of 40.2 kph (25 mph) or less, 56.3 kph (35

mph) or less, and 72.4 kph (45 mph) or less, the fatality rate was 2 percent, 11 percent, and 28

percent, respectively. Additionally, MAIS 1-5 injuries were observed in 6 percent of all rear-end

crashes that occurred on roadways with posted speeds of 40.2 kph (25 mph) or less, 30 percent

with posted speeds of 56.3 kph (35 mph) or less, and 63 percent with posted speeds of 72.4 kph

(45 mph) or less.

b. Crash Imminent Braking (CIB)

If a driver does not take any action to brake when a rear-end crash is imminent, CIB

systems utilize the same types of forward-looking sensors used in DBS systems to apply the

vehicle’s brakes automatically to slow or stop the vehicle. The amount of braking applied varies

by manufacturer, and several systems are designed to achieve maximum vehicle deceleration just

prior to impact. In reviewing model year 2017-2019 NCAP CIB test data, NHTSA observed a

deceleration range of 0.31 to 1.27g during test trials that provided speed reductions capable of

159

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

101

satisfying the CIB performance criteria for a given test condition. Unlike DBS systems, which

only provide additional braking to supplement the driver’s brake input, CIB systems activate

when the driver has not applied the brake pedal.

The Agency’s current CIB test procedure

160

is comprised of the same four test scenarios

(LVS, LVD, LVM, and the STP false positive suppression test) and accompanying test speeds as

set forth in the DBS test procedure. However, the performance criteria vary slightly. The LVM

40.2 kph/16.1 kph (25 mph/10 mph) test condition stipulates that the SV may not contact the

POV. The LVS, LVD, and the LVM 72.4 kph/32.2 kph (45 mph/20 mph) test conditions permit

SV-to-POV contact but require minimum reductions in the SV speed. In the case of the CIB

false positive tests, the performance criterion is little-to-no activation. Similar to NCAP’s DBS

tests, the SSV is the POV presently used in the program’s CIB testing. A short description of

each test scenario and the requirements for a passing result is provided below:

 LVS – SV encounters a stopped POV on a straight road. The SV is moving at 40.2 kph

(25 mph) and the POV (i.e., the SSV) is stationary. The SV throttle is released within

500 ms after the SV issues an FCW alert. To pass this test, the SV speed reduction

attributable to CIB intervention must be ≥ 15.8 kph (9.8 mph).

 LVD – The SV encounters a POV slowing with constant deceleration directly in front of

it on a straight road. The SV and POV are both driven at 56.3 kph (35 mph) with an

initial headway of 13.8 m (45.3 ft.). The POV then decelerates, braking at a constant

deceleration of 0.3g in front of the SV, after which the SV throttle is released within 500

ms after the SV issues an FCW alert. To pass this test, the SV speed reduction

attributable to CIB intervention must be ≥ 16.9 kph (10.5 mph).

160

National Highway Traffic Safety Administration. (2015, October). Crash imminent brake system performance

evaluation for the New Car Assessment Program. http://www.regulations.gov. Docket No. NHTSA-2015-0006-

0025.

102

 LVM – The SV encounters a slower-moving POV directly in front of it on a straight

road. In the first test, the SV and POV are driven on a straight road at a constant speed of

40.2 kph (25 mph) and 16.1 kph (10 mph), respectively. In the second test, the SV and

POV are driven at a constant speed of 72.4 kph (45 mph) and 32.2 kph (20 mph),

respectively. In both tests, the SV throttle is released within 500 ms after the SV issues

an FCW alert. To pass the first test, the SV must not contact the POV. To pass the

second test, the SV speed reduction attributable to CIB intervention must be ≥ 15.8 kph

(9.8 mph).

 STP test (to assess false positive suppression) – The SV is driven towards a steel trench

plate at 40.2 kph (25 mph) in one test and 72.4 kph (45 mph) in the other test. If an FCW

alert is issued, the SV throttle is released within 500 ms of the alert. If no FCW alert is

issued, the throttle is not released until the test’s validity period (the time when all test

specifications and tolerances must be satisfied) has passed. To pass these tests, the SV

must not achieve a peak deceleration equal to or greater than 0.5g at any time during its

approach to the steel trench plate.

To pass NCAP’s CIB system performance criteria, the SV must pass five out of seven

trials for each of the six test conditions.

Similar to FCW and DBS, NCAP’s CIB test scenarios correlate to the dynamically

distinct rear-end crash data discussed earlier. The Agency’s analysis of the 2011-2015 crash data

showed that the LVS, LVD, and LVM scenarios represented 65 percent, 22 percent, and 10

percent, respectively, of all rear-end crashes.

161

With respect to test speed, in its independent

review of 2011-2015 FARS and GES data sets, Volpe concluded that 2 percent of fatal rear-end

crashes and 6 percent of all rear-end crashes occurred on roadways with posted speed limits of

161

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

103

40.2 kph (25 mph) or less. Eleven percent of fatal rear-end crashes and 30 percent of all rear-end

crashes occurred on roads with posted speeds of 56.3 kph (35 mph) or less. For posted speeds of

72.4 kph (45 mph) or less, these statistics are 28 percent and 63 percent, respectively.

c. Current State of AEB Technology

When NHTSA’s CIB test scenarios were developed, relatively few vehicles were

equipped with this technology, and those that were equipped had systems with limited

capabilities. Since then, fitment rates for CIB systems have increased significantly. The

increased fitment was due in part to an industry voluntary commitment made in March 2016. At

that time, 20 vehicle manufacturers, representing more than 99 percent of light motor vehicle

sales in the U.S., voluntarily committed to install AEB systems on light motor vehicles.

162

Pursuant to this voluntary commitment, the manufacturers would make FCW and CIB standard

on virtually all light-duty vehicles with a gross vehicle weight rating (GVWR) of 3,855.5 kg

(8,500 pounds) or less beginning no later than September 1, 2022, and all trucks with a GVWR

between 3,856.0 and 4,535.9 kg (8,501 and 10,000 pounds) beginning no later than September 1,

2025. Conforming vehicles must be equipped with (1) an AEB system that earns at least an

“advanced” rating from IIHS in its front crash prevention track tests and (2) an FCW system that

meets the performance requirements specified in two of NCAP’s three FCW test scenarios.

163

The manufacturers further pledged to submit annual progress reports, which IIHS and NHTSA

agreed to publish. In 2017, the first reporting year, approximately 30 percent of the fleet was

equipped with CIB systems (though many of those systems were not designed to meet the

162

Insurance Institute for Highway Safety (2016, March 17), U.S. DOT and IIHS announce historic commitment of

20 automakers to make automatic emergency braking standard on new vehicles, https://www.iihs.org/news/detail/u-

s-dot-and-iihs-announce-historic-commitment-of-20-automakers-to-make-automatic-emergency-braking-standard-

on-new-vehicles.

163

To achieve an advanced rating in IIHS’ front crash prevention track tests, a vehicle’s AEB system must show a

speed reduction of at least 16.1 kph (10 mph) in either the Institute’s 19.3 or 40.2 kph (12 or 25 mph) tests, or a

speed reduction of 8.0 kph (5 mph) in both of these tests. https://www.iihs.org/news/detail/u-s-dot-and-iihs-

announce-historic-commitment-of-20-automakers-to-make-automatic-emergency-braking-standard-on-new-

vehicles.

104

voluntary commitment thresholds), whereas participating manufacturers equipped 75 percent of

their fleet in 2019.

164

While the voluntary commitment worked to increase fitment rates, the stringency

included in the agreement for AEB systems is lower than that included in NCAP. The voluntary

commitment included front crash prevention track tests that differed in stringency from the

NCAP performance thresholds, and in number. The Agency was aware of those differences at

the time, but considered the voluntary commitment to be a path toward greater fleet

penetration.

165

As fitment has increased, the sensor technology for CIB systems has also advanced

significantly. For instance, in 2017, many systems were not designed to meet the voluntary

commitment thresholds, whereas in 2019, most vehicles with FCW and CIB systems were able

to pass all relevant NCAP test scenarios. NHTSA notes that NCAP’s CIB test requirements

currently require a speed reduction of at least 15.8 kph (9.8 mph) in the program’s LVS test.

These test requirements are more stringent than those required by the voluntary commitment,

which allow a vehicle to comply with the memorandum for a speed reduction of 8.0 kph (5 mph)

in the IIHS 19.3 or 40.2 kph (12 and 25 mph) LVS tests.

166

For the 2021 model year, the pass

rate (as reported by vehicle manufacturers) for NCAP’s FCW and CIB tests for vehicles

167

equipped with these technologies and for which manufacturers submitted data was 88.8 percent

and 69.5 percent, respectively.

168

Furthermore, NHTSA found that 63 percent of model year

164

National Highway Traffic Safety Administration (2019, December 17), NHTSA announces update to historic

AEB commitment by 20 automakers, https://www.nhtsa.gov/press-releases/nhtsa-announces-update-historic-aeb-

commitment-20-automakers.

165

The Agency also believes that its recommendation of AEB systems (i.e., CIB and DBS) that meet NCAP

performance criteria on its website since the 2018 model year has further encouraged adoption of these technologies.

166

Insurance Institute for Highway Safety (2016, March 17), U.S. DOT and IIHS announce historic commitment of

20 automakers to make automatic emergency braking standard on new vehicles, https://www.iihs.org/news/detail/u-

s-dot-and-iihs-announce-historic-commitment-of-20-automakers-to-make-automatic-emergency-braking-standard-

on-new-vehicles.

167

In this instance, “vehicles” refers to the total number of vehicles in the 2021 fleet, and not the total number of

vehicle models for that year.

168

These values assume a fifty percent take rate for vehicles having optional equipment.

105

2017 vehicles did not contact the POV in the LVS scenario during the Agency’s testing, whereas

100 percent of model year 2021 vehicles did not make contact with the POV when tested.

169

such, the Agency believes current CIB system performance far exceeds NCAP’s current testing

requirements, such that it is feasible to update the program’s CIB test conditions to further safety

improvements. Recent NHTSA research supports this assertion.

d. NHTSA’s CIB Characterization Study

Similar to the fleet testing performed for PAEB, the Agency conducted a series of CIB

characterization tests using a sample of MY 2020 NCAP test vehicles from various

manufacturers. The goal of this testing was to quantify the performance of current CIB systems

using the previously defined LVS and LVD test scenarios, but with an expanded set of input

conditions. Testing was conducted in accordance with the CIB test procedure

prescribed above;

however, several scenarios were then repeated to assess how specific procedural changes (i.e.,

increases in test speed and deceleration magnitude) affected CIB system performance.

 For the additional LVS tests, the Agency incrementally increased the vehicle speed for

the LVS test scenario (from 40.2 to 72.4 kph (25 to 45 mph) in 8.0 kph (5 mph)

increments), as shown in Table 2 below, to identify when/if the vehicle reached its

operational limits and/or did not react to the POV ahead. When insufficient intervention

occurred for a given vehicle, the Agency repeated the test scenario at a test speed that

was 4.0 kph (2.5 mph) lower.

170

This reduced speed was used to define the system’s

upper capabilities for the LVS scenario.

 For the additional LVD tests, the Agency evaluated how changes made to either the

vehicles’ speed (72.4 kph versus 56.3 kph (45 mph versus 35 mph)) or deceleration

magnitude (0.5g versus 0.3g) affected CIB performance, as shown in Table 3 below.

169

No contact was assumed if the test vehicle did not contact the POV in 5 or more of the 7 required trial runs.

170

Insufficient intervention was defined as a maximum (peak) deceleration of less than 0.5g.

106

Details of NHTSA’s CIB characterization study are provided below (with speeds given in

kph (mph)):

Table 2: Nominal LVS Matrix

SV Speed, (kph/mph) POV Speed, (kph/mph)

40.2/25 0/0

48.3/30 0/0

56.3/35 0/0

64.4/40 0/0

72.4/45 0/0

Table 3: Nominal LVD Matrix

SV Speed,

(kph/mph)

POV Speed,

(kph/mph)

Peak Deceleration

(g)

Minimum Distance,

(mft.)

56.3/35 56.3/35 0.3 13.8/45.3

56.3/35 56.3/35 0.5 13.8/45.3

72.4/45 72.4/45 0.3 13.8/45.3

No additional LVM or STP false positive assessments were conducted as part of the

Agency’s CIB characterization study. There were several reasons for this. First, in its review of

the 2011-2015 FARS and GES rear-end crash data sets, NHTSA showed that LVS and LVD

rear-end scenarios resulted in the highest number of crashes and MAIS 1-5 injuries. As shown in

Table A-1, there were 1,099,868 LVS, 374,624 LVD, and 174,217 LVM crashes annually.

171

Furthermore, there were 561,842 MAIS 1-5 injuries resulting from the LVS crash scenario,

196,731 for LVD, and 97,402 for LVM. The LVS scenario also had the second highest number

of fatalities. Secondly, it was unclear whether performing a set of additional STP false positive

tests would provide useful data. When the STP test was initially developed, many AEB systems

171

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

107

relied solely on radar for lead vehicle detection. Today, most vehicles utilize camera-only or

fused systems that rely on both camera and radar. Although the Agency has observed instances

of false positive test failures during CIB and DBS NCAP evaluations performed with radar-only

systems, none have been observed when camera-only or fused systems were evaluated in the

program. While some radar-only systems have had difficulty classifying the STP correctly,

camera-only and fused (i.e., camera plus radar) systems have not exhibited this issue.

172

For

these reasons, the Agency believes it may be appropriate to remove the false positive STP

assessments from NCAP’s AEB evaluation matrix in this NCAP update and is seeking comment

in that regard.

The Agency chose to increase the test speeds of the scenarios included in its CIB

characterization study because, in its independent analysis of the 2011-2015 FARS data set,

Volpe found that speeding was a factor in 42 percent of the fatal rear-end crashes.

173

A review

of Volpe’s analysis also showed that approximately 28 percent of fatalities and 63 percent of

injuries in rear-end crashes occurred when the posted speed on roadways is 72.4 kph (45 mph) or

less. When the travel speed was reported in FARS and GES, approximately 7 percent of fatal

and 34 percent of the police reported real-end crashes resulting in injuries occurred at speeds of

72.4 kph (45 mph) or less.

174

These data suggested that there was merit to assessing the

capabilities of newer vehicles using LVS tests performed at higher speeds since this would allow

the Agency to gauge the ability of current-generation CIB systems to address a greater number of

rear-end crashes, particularly those that produce the most serious and fatal injuries. The Agency

also reasoned that it was most appropriate to increase the test speed in NCAP’s LVS scenario, in

particular, since this scenario has the potential to require the greatest speed reduction authority to

172

This is not to suggest that camera systems are superior to radar systems in all tests.

173

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019, August), Statistics of light-

vehicle pre-crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington,

DC: National Highway Traffic Safety Administration.

174

For this crash mode, 62 and 67 percent of the travel speed data is not reported in FARS and GES, respectively.

108

realize potential safety benefits. Historically, it has also been a difficult scenario for forward-

looking sensing systems to address, especially at high vehicle speeds.

Although NHTSA acknowledges that the majority of fatal rear-end crashes (72 percent)

occurred on roads with posted speeds exceeding 72.4 kph (45 mph), these higher speeds were not

assessed as part of the Agency’s characterization testing. Prior to testing, the Agency had safety

concerns with conducting LVS tests at speeds of 80.5 kph (50 mph) or more due to test track

length limitations, inherent safety considerations for laboratory personnel, and potential damage

to either the SV or test equipment. That said, as will be discussed later in this section, data

collected during the Agency’s testing showed that higher test speeds may be feasible, as several

vehicles provided complete crash avoidance at 72.4 kph (45 mph).

NHTSA’s intent in evaluating a modified LVD scenario was to document the

performance of current-generation CIB systems using more demanding LVD-based driving

situations. The Agency also planned to use these test results to determine the feasibility of

increasing the stringency of NCAP’s LVD test. Compared to the LVD test conditions presently

specified in NHTSA’s CIB test procedure, the modified LVD tests, as shown in Table 3, either

(1) maintained the existing 13.8 m (45.3 ft.) SV-to-POV headway and 0.3g POV deceleration

profile, but increased the travel speed of both the POV and SV from 56.3 to 72.4 kph (35 to 45

mph), or (2) maintained the existing 13.8 m (45.3 ft.) SV-to-POV headway and existing 56.3 kph

(35 mph) POV and SV speeds, but increased the average POV deceleration magnitude to 0.5g.

NHTSA’s interest in the first LVD procedural change aligned with that mentioned for the

LVS scenario changes—a significant number of injuries and fatalities in rear-end crashes

occurred at higher speeds. The second change was made to address situations where the driver

of a lead vehicle brakes aggressively, causing the driver of the following vehicle to have even

less time to avoid or mitigate the crash than had the lead vehicle braking been at the 0.3g level

presently specified. The Agency reasoned that implementing these changes for the LVD

109

scenario would introduce a more stringent scenario than that which is currently prescribed in

NHTSA’s CIB test procedure, and would thus help the Agency understand the capabilities of

current CIB systems more comprehensively.

Test reports related to NHTSA’s CIB characterization testing can be found in the docket

for this notice.

e. Updates to NCAP’s CIB Testing

In general, this study has allowed NHTSA to assess the performance of current CIB

systems and evaluate the technology’s future potential for the new model years’ vehicle fleet.

The study showed that many vehicles in today’s fleet were able to repeatedly provide complete

crash avoidance at higher test speeds, shorter SV-to-POV headways, and generally more

aggressive conditions than those specified in the Agency’s current NCAP CIB test procedure.

This study has also provided the Agency with new ways to consider differentiating CIB systems’

performance for NCAP ratings purposes in the future. Furthermore, it has provided the Agency

with the underlying support necessary for NCAP to propose adjustments to the current CIB

performance requirements to address rear-end crashes that are causing a greater number of

injuries and fatalities in the real world. Accordingly, the Agency is proposing to make several

changes to its CIB test procedure for this NCAP upgrade. These changes are outlined below for

each test scenario. For the LVS scenario, the Agency is proposing the following:

 Increased SV test speeds and an assessment methodology that is similar to that which it

proposed to assess PAEB system performance. CIB system performance for the LVS

scenario will be assessed over a range of test speeds. The Agency is proposing a

minimum SV test speed of 40 kph (24.9 mph), which is similar to that currently specified

in NHTSA’s CIB test procedure – 40.2 kph (25 mph), and a maximum SV test speed of

80.0 kph (49.7 mph). The Agency is proposing to increase the subject vehicle test speed

110

in 10 kph (6.2 mph) increments from the minimum test speed to the maximum test speed

for the LVS assessment.

The Agency’s characterization testing showed that it is feasible to raise the SV

speed in NCAP’s LVS test to encourage improved performance of CIB systems. In fact,

several vehicles repeatably afforded full crash avoidance (i.e., no contact) at speeds up to

72.4 kph (45 mph) for the LVS test scenario. Furthermore, NHTSA recognizes that Euro

NCAP performs its Car-to-Car Rear stationary (CCRs) scenario, which is comparable to

the Agency’s LVS tests, at speeds as high as 80 kph (49.7 mph) for those systems that

offer AEB, which also suggests that higher test speeds are practicable.

175

As such,

NHTSA believes that it is appropriate to harmonize with Euro NCAP on the maximum

LVS test speed of 80 kph (49.7 mph), as this should better address the higher severity,

high-speed crash problem and, in turn, further reduce fatalities and serious injuries.

Although Euro NCAP’s protocol prescribes a minimum SV test speed of 10 kph (6.2

mph) for the CCRs scenario for AEB systems that also offer FCW, the Agency does not

see a reason to perform its LVS test at a speed that is less than that which is specified in

its existing test procedure (40.2 kph (25 mph)). Therefore, it is not proposing to

harmonize with Euro NCAP with respect to the minimum required test speed.

 A revised performance requirement. In lieu of a speed reduction, as is currently specified

in NHTSA’s CIB test procedure for the LVS scenario, the SV must avoid making contact

with the POV target to pass a test trial. Similar to PAEB, this should limit damage to the

SV and POV target during testing and reduce chances that results are questioned or

invalidated.

175

European New Car Assessment Programme (Euro NCAP) (April 2021), Test Protocol – AEB Car-to-Car

systems, Version 3.0.3. See section 8.2.3.

111

 Changes to the number of test trials required for the LVS scenario. Currently, NHTSA’s

CIB test procedure requires that a vehicle meet the performance criteria (i.e., specified

speed reduction) for five out of seven trials. However, similar to that proposed by

NHTSA for its PAEB assessment, the Agency is proposing that only one test trial will be

conducted per test speed assessed (i.e., 40, 50, 60, 70, and 80 kph or 24.9, 31.1, 37.3,

43.5, and 49.7 mph) if the SV does not contact the POV target during the first valid trial

for each of the test speeds. For a given test condition, the test sequence is initiated at the

40 kph (24.9 mph) minimum speed. To achieve a passing result, the test must be valid

(i.e., all test specifications and tolerances satisfied), and the SV must not contact the

POV. If the SV does not contact the POV during the first valid test, the test speed is

incrementally increased by 10 kph (6.2 mph), and the next test in the sequence is

performed. Unless the SV contacts the POV, this iterative process continues until a

maximum test speed of 80 kph (31.1 mph) is evaluated. If the SV contacts the POV, and

the relative longitudinal velocity between the SV and POV is less than or equal to 50

percent of the initial speed of the SV, the Agency will perform four additional (repeated)

test trials at the same speed for which the impact occurred. The SV must not contact the

POV for at least three out of the five test trials performed at that same speed to pass that

specific combination of test condition and test speed.

176

If the SV contacts the POV

during a valid test of a test condition (whether it be the first test performed for a

particular test speed or a subsequent test trial at that same speed), and the relative impact

velocity exceeds 50 percent of the initial speed of the SV, no additional test trials will be

conducted at the given test speed and test condition and the SV is considered to have

failed the test condition at that specific test speed.

176

The Agency notes that a similar pass/fail criterion (i.e., a vehicle must meet performance requirements for three

out of five trials for a particular test condition to pass the test condition) is included in its LDW test procedure, as

referenced earlier.

112

The Agency is pursuing an assessment approach for the LVS CIB test scenario

that is similar to that proposed for PAEB systems in order to reduce test burden, given

that additional test speeds are being proposed. NHTSA believes that this alternative

approach will continue to ensure that passing CIB systems represent robust designs that

will offer a higher level of performance and safety.

For the LVD scenario, the Agency is proposing the following:

 A reduction in SV and POV test speeds. NHTSA’s CIB test procedure currently

prescribes a test speed of 56 kph (34.8 mph) for the SV and POV in the LVD scenario.

Euro NCAP’s AEB Car-to-Car systems test protocol, Version 3.0.3, dated April 2021 for

the Car-to-Car rear braking (CCRb) specifies an SV speed of 50 kph (31.1 mph). For this

upgrade of NCAP, the Agency is proposing to reduce the test speed for the SV and POV

to 50 kph (31.1 mph) to harmonize with Euro NCAP.

177

Given additional changes

proposed for the SV-to-POV headway and deceleration magnitude (discussed next),

NHTSA does not believe the proposed reduction in test speed will lead to an overall

reduction in test stringency or loss of safety benefits.

The Agency is also requesting comment on whether it is appropriate to

incorporate additional SV test speeds for the LVD test scenario, specifically 60, 70, and

80 kph (37.3, 43.5, and 49.7 mph) or, alternatively, whether testing at only 50 kph (31.1

mph) and 80 kph (49.7 mph) would be sufficient. As mentioned earlier, Volpe’s analysis

of the 2011-2015 FARS data set showed that the majority of crashes occurred on roads

with posted speeds exceeding 72.4 kph (45 mph), suggesting that testing at higher speeds

for all CIB test scenarios may be warranted. The Agency has simply not performed

testing at 80 kph (49.7 mph) to date because of concerns surrounding laboratories’

177

European New Car Assessment Programme (Euro NCAP) (April 2021), Test Protocol – AEB Car-to-Car

systems, Version 3.0.3. See section 8.2.5.

113

abilities to safely execute such tests and limited available testing real estate, as this test

scenario requires that both the SV and POV be travelling at the same speed at the onset of

the test validity period. That being said, NHTSA believes that, 1) given the results from

its characterization study, and in particular, the braking performance demonstrated in the

LVS tests, 2) the fact that tested vehicles may have higher POV classification confidence

for the LVD test compared to the LVS test since the POV is always in motion during the

LVD test, and 3) the POV will be the GVT, which relies on a robotic platform for

movement, rather than the SSV which must be towed along a monorail secured to the test

track, vehicles in the current fleet will likely also perform well in higher speed LVD tests.

To validate this assumption, NHTSA will be conducting research next year to assess

vehicle performance at speeds ranging from 50 kph (31.1 mph) to 80 kph (49.7 mph) for

12 and 40 m (39.4 and 131.2 ft.) headways and POV deceleration magnitudes of 0.4 and

0.5 g for the LVD CIB test scenario. Pending the outcome of that research, the Agency

may consider adopting additional higher tests speeds (i.e., 60, 70, and/or 80 kph (37.3,

43.5, and/or 49.7 mph)) for the LVD test scenario in NCAP. The Agency requests

comment on what SV-to-POV headway and deceleration magnitude(s) would be

appropriate if the Agency was to adopt any or all of these additional test speeds. If

additional test speeds are adopted, the Agency would implement an assessment

methodology similar to that proposed for the CIB LVS test scenario, whereby NHTSA

would increase the SV test speed in 10 kph (6.2 mph) increments from the minimum test

speed to the maximum test speed for the LVD assessment.

 A reduction in SV-to-POV headway. NHTSA’s CIB test procedure currently specifies a

13.8 m (45.3 ft.) SV-to-POV headway for the LVD scenario. The Agency is proposing to

reduce the prescribed headway to 12 m (39.4 ft.) to harmonize with Euro NCAP’s CCRb

scenario. Given the proposed test speed reduction, the Agency believes it is appropriate

114

to also reduce the headway to maintain similar stringency with its current LVD test

condition. Whereas Euro NCAP also specifies an additional SV-to-POV headway of 40

m (131.2 ft.), the Agency is not proposing to conduct this additional assessment as part of

this proposal. NHTSA does not believe there would be a safety benefit to adopting 40 m

(131.2 ft.) as an additional, and less stringent, headway. Therefore, it would serve to

increase the test burden unnecessarily.

 An increase in deceleration magnitude. The Agency is proposing to increase the POV

deceleration magnitude currently specified in its CIB test procedure for the LVD scenario

from 0.3 g to 0.5 g. In the Agency’s CIB characterization study, some vehicles

repeatably afforded full crash avoidance (i.e., no contact) for all trials when the POV

executed a 0.5 g braking maneuver in the LVD condition with a SV test speed of 35 mph

and SV-to-POV headway of 13.8 m (45.3 ft.). Although the test speed used in the

Agency’s study was slightly lower than that which the Agency is proposing for the LVD

test condition, and the SV-to-POV headway was slightly longer, NHTSA believes that it

is reasonable to adopt a higher POV deceleration magnitude for its future LVD testing.

The Agency notes that a deceleration of 0.5 g falls within the range of deceleration

magnitudes prescribed by Euro NCAP in its AEB Car-to-Car systems test protocol,

Version 3.0.3, dated April 2021 for the CCRb scenario. In its CCRb test, Euro NCAP

specifies POV deceleration magnitudes of 2 m/s

and 6 m/s

(approximately 0.2 to 0.6 g)

for an SV-to-POV headway of 12 m (39.4 ft.) and SV test speed of 50 kph (31.1 mph).

As the Agency has proposed this reduced headway and test speed for its LVD testing, it

reasons that adopting a 0.5 g POV deceleration magnitude is also practicable. The

Agency is not proposing 0.6 g as the POV deceleration magnitude in its LVD test

because it has observed instances where the tires on the POV target developed flat spots

115

during research testing conducted with the Guided Soft Target (GST) system

178

to assess

Traffic Jam Assist (TJA) systems. The TJA testing required a braking maneuver for the

lead vehicle decelerates, accelerates, then decelerates (LVDAD) scenario that is similar

to that specified in the Agency’s CIB LVD test.

179

During this testing, NHTSA also

found that it was more difficult to achieve and accurately control deceleration when

braking maneuvers higher than 0.5 g were used.

180

Extensive tuning efforts related to the

GST brake applications were made in an attempt to rectify the problems encountered, but

these adjustments were unable to consistently satisfy the test tolerances associated with

0.6 g POV deceleration for the LVDAD test and a recommendation was made to reduce

the maximum nominal POV deceleration from 0.6 g to 0.5 g for future testing. In its

report findings, the Agency also noted that a deceleration of 0.6 g is not only very close

to the maximum braking capability of the GST’s robotic platform used by the Agency, it

is also very close to the default magnitude used by the LPRV during an emergency stop

(maximum deceleration). As such, the Agency concluded that a decrease in maximum

POV deceleration should also reduce equipment wear, particularly for the system’s tires

and braking components, thus improving test efficiency. This being said, the Agency

acknowledges that newer robotic platforms designed to provide greater capabilities, are

now becoming available, which may resolve the issues observed in the Agency’s TJA

testing. As such, the Agency is requesting comment on whether it is feasible to adopt a

POV deceleration magnitude of 0.6 g in lieu of 0.5 g, as proposed.

178

The GST system is comprised of two main parts – a low profile robotic vehicle (LPRV), and a global vehicle

target (GVT), which is secured to the top of the LPRV.

179

Fogle, E. E., Arquette, T. E. (TRC), and Forkenbrock, G. J. (NHTSA), (2021, May), Traffic Jam Assist Draft Test

Procedure Performability Validation (Report No. DOT HS 812 987), Washington, DC: National Highway Traffic

Safety Administration.

180

From Section 4.1 of DOT HS 812 987 – “POV deceleration validity check failures occurred during six trials of

the eight LVDAD trials performed. Four of the seven 0.6 g failures were because the POV was unable to achieve

the minimum deceleration threshold of 0.55 g. The remaining three 0.6 g failures were because the POV was unable

to maintain a minimum average deceleration of at least 0.55 g.”

116

 An alternative performance criterion. In lieu of a speed reduction, as is currently

specified in NHTSA’s CIB test procedure for the LVD scenario, the vehicle must avoid

making contact with the POV target to pass a test trial.

 Changes to the number of test trials required for the LVD scenario. NHTSA is adopting

an approach to conducting test trials that is identical to that described above for the CIB

LVS scenario, regardless of the number of test speeds adopted (i.e., one speed, 50 kph

(31.1 mph); two speeds, 50 kph (31.1 mph) and 80 kph (49.7 mph); or four speeds, 50,

60, 70, and 80 kph (31.1, 37.3, 43.5, and 49.7 mph)). If only one or two test speeds are

selected for inclusion, the Agency is seeking comment on whether it is more appropriate

to alternatively require 7 trials for each test speed, and require that 5 out of the 7 trials

conducted pass the “no contact” performance criterion.

For the LVM scenario, the Agency is proposing the following:

 Increased SV test speeds. NHTSA is proposing to assess CIB system performance for

the LVM scenario over a range of test speeds, similar to that proposed for the LVS

scenario. The Agency is proposing a minimum SV test speed of 40 kph (24.9 mph),

which is nearly equivalent to the 40.2 kph (25 mph) test speed currently specified in

NHTSA’s CIB test procedure, and a maximum SV test speed of 80 kph (49.7 mph),

which is slightly higher than the 72.4 kph (45 mph) specified for the second LVM test

condition in NHTSA’s current CIB test procedure. The Agency is proposing to increase

the SV test speed in 10 kph (6.2 mph) increments from the minimum test speed to the

maximum test speed for the LVM assessment.

The Agency did not perform additional LVM testing as part of its CIB

characterization study. Nonetheless, NHTSA believes that it is feasible to raise the SV

speed in NCAP’s LVM test to encourage improved performance of CIB systems, as the

Agency’s current CIB LVM tests (conducted with an SV speed of 72.4 kph (45 mph) and

117

POV speed of 32.2 kph (20 mph)) have shown that many vehicles are able to stop

without contacting the POV target for each of the required test trials. Furthermore,

NHTSA recognizes that Euro NCAP performs its Car-to-Car Rear moving (CCRm)

scenario, which is comparable to the Agency’s LVM tests, at speeds as high as 80 kph

(49.7 mph), which also suggests that higher SV test speeds are practicable.

181

As such,

NHTSA believes that it is appropriate to harmonize with Euro NCAP on the maximum

SV test speed of 80 kph (49.7 mph) in the Agency’s LVM test, as this should also address

high-speed crashes and thus further reduce fatalities and serious injuries. Although Euro

NCAP’s protocol prescribes a minimum SV test speed of 30 kph (18.6 mph) for the

CCRm scenario for vehicles that have AEB systems,

182

the Agency does not see a reason

to perform its LVM test at a speed that is less than that which is specified in its existing

test procedure (40.2 kph (25 mph)). Therefore, it is not proposing to harmonize with

Euro NCAP with respect to the minimum required test speed.

 An alternative POV test speed for all test conditions. While the Agency’s CIB test

procedure currently specifies a POV test speed of 16.1 kph (10 mph) when the SV speed

is 40.2 kph (25 mph) and a POV test speed of 32.2 kph (20 mph) when the SV speed is

72.4 kph (45 mph), the Agency is proposing to use a POV test speed of 20 kph (12.4

mph) for every SV test speed that will be assessed for the LVM scenario; 40 to 80 kph

(24.9 to 49.7 mph), increased in 10.0 kph (6.2 mph) increments. NHTSA recognizes that

Euro NCAP’s CCRm protocol specifies a POV test speed of 20 kph (12.4 mph), and this

POV speed is stipulated for similar testing conducted by various other vehicle safety

181

European New Car Assessment Programme (Euro NCAP) (April 2021), Test Protocol – AEB Car-to-Car

systems, Version 3.0.3. See section 8.2.3.

182

The Agency notes that the minimum SV test for vehicles equipped with only FCW (and no AEB) is 50 kph (31.1

mph).

118

ratings programs. With this proposed NCAP upgrade, NHTSA sees no reason to deviate

from the other testing organizations with respect to the POV speed for its LVM test.

 A performance criterion of “no contact”. In lieu of a speed reduction, as is currently

specified in NHTSA’s CIB test procedure for the Agency’s higher speed LVM scenario

(i.e., POV of 72.4 kph (45 mph) and POV speed of 32.2 kph (20 mph)), the SV must

avoid making contact with the POV target to pass a test trial for each test speed assessed

for the LVM scenario; 40 to 80 kph (24.9 to 49.7 mph), increased in 10 kph (6.2 mph)

increments.

 Changes to the number of test trials required for the LVM scenario. NHTSA is adopting

an approach to conducting test trials that is identical to that described above for the CIB

LVS scenario. For the proposed CIB LVM tests, the Agency would require one test trial

per SV speed increment, and four repeat trials in the event of a test failure for instances

where the SV has a relative velocity at impact that is equal to or less than 50 percent of

the initial speed.

NHTSA has chosen to harmonize with Euro NCAP in many respects since it recognizes

that the rear-end crash problem, as defined by the most frequently occurring and dynamically

distinct pre-crash scenarios, could be changing as AEB-equipped vehicles become more prolific

in the fleet. Accordingly, the Agency believes that it is beneficial to standardize the current CIB

test specifications with other consumer information programs and focus resources on emerging

trends.

183

However, the Agency also notes that it will consider making additional updates to its

CIB test evaluation as the crash problem evolves.

f. Updates to NCAP’s DBS Testing

183

Cicchino, J. B. & Zuby, D. S. (2019, August), Characteristics of rear-end crashes involving passenger vehicles

with automatic emergency braking, Traffic Injury Prevention, 2019, VOL. 20, NO. S1, S112–S118,

https://doi.org/10.1080/15389588.2019.1576172.

119

NHTSA did not conduct any testing, as part of its characterization study, to evaluate DBS

system performance capabilities beyond what is currently stipulated in NCAP’s DBS test

procedure. However, the Agency notes that its CIB and DBS test procedures are currently

aligned with respect to test scenarios, test speeds, headways, etc. Differences exist only with

respect to the use of an SV manual brake application (i.e., for DBS) and most performance

criterion. NHTSA’s DBS test procedure currently specifies “no contact” as the performance

criterion for all DBS test conditions, whereas the Agency’s CIB test procedure currently requires

a specified speed reduction for each of the CIB test conditions (with the exception of the lower

speed LVM condition where the POV speed is 16.1 kph (10 mph) and the SV speed is 40.2 kph

(25 mph), which requires “no contact”). Therefore, NHTSA believes it is reasonable to adopt the

CIB test conditions (i.e., test speeds, headways, etc.) for the comparable DBS test

conditions. However, given the Agency’s proposal to embrace the more stringent “no contact”

performance criterion for each of the CIB test conditions, and for the additional reasons

mentioned previously, the Agency also believes, as suggested prior, that there may be merit to

removing the DBS test conditions from NCAP entirely to reduce test burden and the associated

cost.

In its comments to the NCAP’s December 2015 notice, the Alliance

184

stated that since

crash avoidance (i.e., no vehicle contact) is the desired outcome for all imminent rear-end crash

events, if an SV avoids contact with the POV in all CIB tests, DBS testing should not be

necessary. Although NHTSA agrees with the Alliance’s rationale in principle, the Agency also

believes there is merit to ensuring that both AEB systems perform as designed and help the

driver to mitigate or prevent the crash. The Agency reasons that it is possible for the driver to

apply the brakes, but with a magnitude that does not result in achieving the vehicle’s maximum

184

The Agency notes that the Alliance of Automobile Manufacturers (The Alliance) merged with Global

Automakers in January 2020 to create the Alliance for Automotive Innovation (Auto Innovators). Both automotive

industry groups separately submitted comments to the December 2015 notice.

120

crash avoidance potential (i.e., deceleration). In the past, some manufacturers assumed the driver

was in control when the brake pedal was depressed and would not override the driver’s input

when necessary to avoid a crash. Accordingly, NHTSA hesitates to assume that if CIB systems

work effectively during testing, then DBS systems will automatically do so as well.

In light of these considerations, the Agency is tentatively proposing to retain both CIB

and DBS system performance tests in NCAP, and to align all test conditions for comparable test

scenarios (e.g., SV and POV test speeds, headway, etc.) to evaluate whether the DBS system will

provide supplemental braking if the driver brakes but additional braking is warranted. For this

testing, the Agency is proposing to adopt an assessment approach for DBS that is identical to that

described previously for PAEB and CIB. The Agency would require one test trial per speed for

each test scenario, and four repeated trials for any specific test condition and speed combination

that results in a test failure and where the SV has a relative velocity at impact that is equal to or

less than 50 percent of the initial speed. Speeds will be increased in 10 kph (6.2 mph)

increments from the minimum test speed to the maximum test speed. However, the Agency is

also requesting comment on whether removal of the DBS test scenarios from NCAP would be

more appropriate.

As an alternative to retaining all DBS tests in NCAP, or removing the DBS performance

evaluations from NCAP entirely, the Agency believes it may be more reasonable to conduct only

the LVS and LVM tests at the highest two test speeds proposed for CIB – 70 and 80 kph (43.5

and 49.7 mph) – to ensure system functionality and that the SV will not suppress AEB operation

when the driver applies the vehicle’s foundation brakes. The Agency would also consider

conducting the LVD DBS test at 70 and 80 kph (43.5 and 49.7 mph) if the Agency decides to

also adopt these test speeds for the related CIB test. Comments are requested on this alternative

proposal and whether an alternative assessment method would be more appropriate if any or all

of the DBS test scenarios were conducted only at the two highest test speeds. For a more limited

121

speed assessment of the two highest test speeds, 70 and 80 kph (43.5 and 49.7 mph), instead of

up to four test speeds (50, 60, 70, and 80 kph (31.1, 37.3, 43.5, and 49.7 mph)) for LVD, or five

test speeds (40, 50, 60, 70, and 80 kph (24.9, 31.1, 37.3, 43.5, and 49.7 mph)) for LVS and

LVM), should the Agency require one trial per test condition (i.e., align with the assessment

method outlined for the other AEB test conditions) or multiple trials? If multiple trials were to

be required, how many would be appropriate, and what would be an acceptable pass rate?

If the Agency continues to perform DBS testing in NCAP, it also proposes to revise when

the manual (robotic) brake application is initiated. The current DBS test procedure prescribes

this shall occur at specific TTCs per test scenario: 1.1 seconds (LVS), 1.0 seconds (LVM), and

1.4 second (LVD). The proposed revision would initiate manual braking at a time that

corresponds to 1.0 second after the FCW alert is issued for all DBS test scenario and speed

combinations, regardless of whether a CIB activation occurs after the FCW alert but before

initiation of the manual brake application. The Agency reasons that this change is more

representative of real-world use and driving conditions, and is in basic agreement with the

approach specified for FCW performance evaluations in Euro NCAP’s AEB Car-to-Car systems

test protocol.

185

Alternatively, the Agency requests comment on appropriate TTCs for the

modified test conditions.

g. Updates to NCAP’s FCW Testing

As mentioned earlier, NHTSA is proposing to consolidate its FCW and CIB tests such

that the CIB tests will be used as an indicant of FCW operation. The Agency is also proposing to

similarly assess FCW in the context of its PAEB tests. NHTSA believes there is merit to

assessing the presence of an FCW alert within the CIB and PAEB test because operation of FCW

and AEB/PAEB systems, in the test scenarios to be used by NCAP, are complementary and

185

European New Car Assessment Programme (Euro NCAP) (April 2021), Test Protocol – AEB Car-to-Car

systems, Version 3.0.3. See Annex A.

122

fundamentally intertwined. Also, combining the Agency’s FCW tests with those used to assess

AEB system performance would reduce test burden. The Agency proposes that it would

evaluate the presence of a vehicle’s FCW system during its CIB tests by requiring the SV

accelerator pedal be fully released within 500 ms after the FCW alert is issued. If no FCW alert

is issued during a CIB test, the SV accelerator pedal will be fully released within 500 ms after

the onset of CIB system braking.

186

Here, the onset of CIB activation is taken to be the instant

SV deceleration reaches at least 0.5g. If no FCW alert is issued and the vehicle’s CIB system

does not offer any braking, release of the SV accelerator pedal will not be required prior to

impact with the POV. The Agency is also proposing to make similar procedural changes to its

PAEB test procedure. NHTSA is seeking comment as to whether the proposed FCW assessment

method is reasonable. Furthermore, given that most FCW systems are currently able to pass all

relevant NCAP test scenarios, as mentioned earlier, the Agency believes that, as an alternative to

integrating the assessment of FCW into the Agency’s CIB tests, it may be feasible for NCAP to

perform one FCW test that could serve as an indicant of FCW system performance (while still

retaining the previously-stated accelerator pedal release timing to ensure CIB activation is not

unintentionally suppressed). This would also reduce test burden. If the Agency were to choose

one of the proposed CIB test scenarios to adopt for an FCW test to assess the performance of

FCW systems, which CIB test scenario do commenters believe would be most appropriate and

why?

The Agency notes that if it maintains any or all of the FCW test scenarios that are

currently included in its FCW test procedure, it proposes to align the corresponding maximum

SV test speeds, POV speeds, headway, POV deceleration magnitude, etc., as applicable, with the

186

Previous NHTSA research indicates that human drivers are capable of releasing the accelerator pedal within 500

ms after returning their eyes to a forward-facing viewing position in response to an FCW alert. Forkenbrock, G.,

Snyder, A., Hoover, R., O’Harra, B., Vasko, S., Smith, L. (2011, July), A Test Track Protocol for Assessing

Forward Collision Warning Driver-Vehicle Interface Effectiveness (Report No. DOT HS 811 501), Washington,

DC: National Highway Traffic Safety Administration.

123

included CIB tests, similar to that which it has proposed for the DBS tests. Accordingly, the

Agency would adopt the following for the FCW tests:

 LVS – SV speed of 80 kph (49.7 mph); POV is stationary.

 LVD – SV and POV speed of 50 kph (31.1 mph) or up to 80 kph (49.7 mph), depending

on the final test speed adopted for the CIB LVD scenario; a 12 m (39.4 ft.) SV-to-POV

headway; and a POV deceleration magnitude of 0.5 g.

 LVM – SV speed of 80 kph (49.7 mph); POV speed of 20 kph (12.4 mph).

If the Agency continues to conduct separate FCW assessments, it will need to revise the

prescribed TTCs currently used to assess FCW performance to align with the revised test

scenario and speed combinations.

187

Given the Agency’s thoughts about FCW-AEB integration

and the revised test conditions that would be adopted for any future FCW tests, NHTSA requests

comment on what TTC would be appropriate for each test scenario. Although the Agency is

proposing to adopt an assessment approach for FCW that is identical to that described previously

for PAEB, CIB, and DBS,

188

it is also requesting comment on whether an alternative assessment

method would be appropriate in instances where it retains one or more FCW scenarios that are

performed at a single test speed. In such instances, should the Agency require one trial per test

condition (i.e., align with the assessment method outlined for the other AEB test conditions) or

multiple trials? If multiple trials were to be required, how many would be appropriate, and what

would be an acceptable pass rate?

h. Regenerative Braking

187

To pass a test trial, the vehicle must issue the FCW alert on or prior to the prescribed time-to-collision (TTC)

specified for each of the three FCW test scenarios.

188

In essence, the Agency would require one test trial per speed for each test scenario and four repeat trials in the

event of a test failure for instances where the SV has a relative velocity at impact that is equal to or less than 50

percent of the initial speed. Speeds will be increased in 10 kph (6.2 mph) increments from the minimum test speed

to the maximum test speed.

124

In addition to the FCW alert setting, discussed earlier, there are additional system settings

that the Agency must now consider during its AEB and PAEB testing. One such setting is that

for regenerative braking. Regenerative braking, which has become more common as electric

vehicles have begun to proliferate the fleet, can slow the vehicle when the throttle is released.

As such, when the throttle is fully released upon the issuance of the FCW alert in the Agency’s

AEB and PAEB testing, vehicle speed can reduce significantly prior to the onset of braking

associated with these technologies, particularly in instances where the FCW alert is issued early.

For vehicles with regenerative braking that have multiple settings (e.g., nominal, more

aggressive, less aggressive), the Agency is proposing to use the “off” setting or the setting that

provides the lowest deceleration when the accelerator is fully released in its AEB and PAEB

tests.

189

Although NHTSA reasons that the nominal setting may be the setting most commonly

chosen by a typical driver, it prefers the least aggressive setting, as it would be more indicative

of “worst case”. Selecting a setting that affords the lowest deceleration allows the vehicle to

travel faster at the onset of braking associated with AEB and PAEB. This approach would

produce a situation that is more comparable to that for vehicles that do not have regenerative

braking.

The Agency believes that regenerative braking may also introduce complications for the

Agency’s DBS tests (if the DBS tests are retained in NCAP). NHTSA reasons that some

vehicles may offer regenerative braking that is already so high that there would be only a

relatively small boost in braking from the braking actuator (acting to provide a combined 0.4 g

deceleration). For instance, if the regenerative braking from simply releasing the accelerator

pedal results in 0.3 g braking, the additional braking required to get to 0.4 g from the actuator

would be a very low force and/or brake pedal displacement. The Agency is requesting comment

189

The Agency does not plan to make any procedural modifications for vehicles that have regenerative braking that

cannot be switched off or adjusted, as those vehicles should operate similarly in the real world.

125

on whether regenerative braking may introduce additional testing issues and on any

recommendations for test procedural changes to rectify possible testing issues related to

regenerative braking.

With respect to FCW, CIB, and DBS testing in NCAP, NHTSA is seeking comment on

the following:

(38) For the Agency’s FCW tests:

- If the Agency retains one or more separate tests for FCW, should it award credit

solely to vehicles equipped with FCW systems that provide a passing audible

alert? Or, should it also consider awarding credit to vehicles equipped with FCW

systems that provide passing haptic alerts? Are there certain haptic alert types

that should be excluded from consideration (if the Agency was to award credit to

vehicles with haptic alerts that pass NCAP tests) because they may be a nuisance

to drivers such that they are more likely to disable the system? Do commenters

believe that haptic alerts can be accurately and objectively assessed? Why or why

not? Is it appropriate for the Agency to refrain from awarding credit to FCW

systems that provide only a passing visual alert? Why or why not? If the Agency

assesses the sufficiency of the FCW alert in the context of CIB (and PAEB) tests,

what type of FCW alert(s) would be acceptable for use in defining the timing of

the release of the SV accelerator pedal, and why?

- Is it most appropriate to test the middle (or next latest) FCW system setting in lieu

of the default setting when performing FCW and AEB (including PAEB) NCAP

tests on vehicles that offer multiple FCW timing adjustment settings? Why or

why not? If not, what use setting would be most appropriate?

- Should the Agency consider consolidating FCW and CIB testing such that

NCAP’s CIB test scenarios would serve as an indicant of FCW operation? Why

126

or why not? The Agency has proposed that if it combines the two tests, it would

evaluate the presence of a vehicle’s FCW system during its CIB tests by requiring

the SV accelerator pedal be fully released within 500 ms after the FCW alert is

issued. If no FCW alert is issued during a CIB test, the SV accelerator pedal will

be fully released within 500 ms after the onset of CIB system braking (as defined

by the instant SV deceleration reaches at least 0.5g). If no FCW alert is issued

and the vehicle’s CIB system does not offer any braking, release of the SV

accelerator pedal will not be required prior to impact with the POV. The Agency

notes that it has also proposed these test procedural changes for its PAEB tests as

well. Is this assessment method for FCW operation reasonable? Why or why

not?

- If the Agency continues to assess FCW systems separately from CIB, how should

the current FCW performance criteria (i.e., TTCs) be amended if the Agency

aligns the corresponding maximum SV test speeds, POV speeds, SV-to-POV

headway, POV deceleration magnitude, etc., as applicable, with the proposed CIB

tests, and why? What assessment method should be used – one trial per scenario,

or multiple trials, and why? If multiple trials should be required, how many

would be appropriate, and why? Also, what would be an acceptable pass rate, and

why?

- Is it desirable for NCAP to perform one FCW test scenario (instead of the three

that are currently included in NCAP’s FCW test procedure), conducted at the

corresponding maximum SV test speed, POV speed, SV-to-POV headway (as

applicable), POV deceleration magnitude, etc. of the proposed CIB test to serve as

an indicant of FCW system performance? If so, which test scenario from NCAP’s

FCW test procedure is appropriate?

127

- Are there additional or alternative test scenarios or test conditions that the Agency

should consider incorporating into the FCW test procedure, such as those at even

higher test speeds than those proposed for the CIB tests, or those having increased

complexity? If so, should the current FCW performance criteria (i.e., TTCs)

and/or test scenario specifications be amended, and to what extent?

(39) For the Agency’s CIB tests:

- Are the SV and POV speeds, SV-to-POV headway, deceleration magnitude, etc.

the Agency has proposed for NCAP’s CIB tests appropriate? Why or why not? If

not, what speeds, headway(s), deceleration magnitude(s) are appropriate, and

why? Should the Agency adopt a POV deceleration magnitude of 0.6 g for its

LVD CIB test in lieu of 0.5 g proposed? Why or why not?

- Should the Agency consider adopting additional higher tests speeds (i.e., 60, 70,

and/or 80 kph (37.3, 43.5, and/or 49.7 mph)) for the CIB (and potentially DBS)

LVD test scenario in NCAP? Why or why not? If additional speeds are included,

what headway and deceleration magnitude would be appropriate for each

additional test speed, and why?

- Is a performance criterion of “no contact” appropriate for the proposed CIB and

DBS test conditions? Why or why not? Alternatively, should the Agency require

minimum speed reductions or specify a maximum allowable SV-to-POV impact

speed for any or all of the proposed test conditions (i.e., test scenario and test

speed combination)? If yes, why, and for which test conditions? For those test

conditions, what speed reductions would be appropriate? Alternatively, what

maximum allowable impact speed would be appropriate?

(40) For the Agency’s DBS tests:

128

- Should the Agency remove the DBS test scenarios from NCAP? Why or why

not? Alternatively, should the Agency conduct the DBS LVS and LVM tests at

only the highest test speeds proposed for CIB – 70 and 80 kph (43.5 and 49.7

mph)? Why or why not? If the Agency also adopted these higher tests speeds (70

and 80 kph (43.5 and 49.7 mph)) for the LVD CIB test, should it also conduct the

LVD DBS test at these same speeds? Why or why not?

- If the Agency continues to perform DBS testing in NCAP, is it appropriate to

revise when the manual (robotic) brake application is initiated to a time that

corresponds to 1.0 second after the FCW alert is issued (regardless of whether a

CIB activation occurs after the FCW alert but before initiation of the manual

brake application)? If not, why, and what prescribed TTC values would be

appropriate for the modified DBS test conditions?

(41) Is the assessment method NHTSA has proposed for the CIB and DBS tests (i.e., one

trial per test speed with speed increments of 10 kph (6.2 mph) for each test condition

and repeat trials only in the event of POV contact) appropriate? Why or why not?

Should an alternative assessment method such as multiple trials be required instead? If

yes, why? If multiple trials should be required, how many would be appropriate, and

why? Also, what would be an acceptable pass rate, and why? If the proposed

assessment method is appropriate, it is acceptable even for the LVD test scenario if

only one or two test speeds are selected for inclusion? Or, is it more appropriate to

alternatively require 7 trials for each test speed, and require that 5 out of the 7 trials

conducted pass the “no contact” performance criterion?

(42) The Agency’s proposal to (1) consolidate its FCW and CIB tests such that the CIB tests

would also serve as an indicant of FCW operation, (2) assess 14 test speeds for CIB (5

for LVS, 5 for LVM, and potentially 4 for LVD), and (3) assess 6 tests speeds for DBS

129

(2 for LVS, 2 for LVM, and potentially 2 for LVD), would result in a total of 20 unique

combinations of test conditions and test speeds to be evaluated for AEB. If the Agency

uses check marks to give credit to vehicles that (1) are equipped with the recommended

ADAS technologies, and (2) pass the applicable system performance test requirements

for each ADAS technology included in NCAP until such time as a new ADAS rating

system is developed and a final rule to amend the safety rating section of the Monroney

label is published, what is an appropriate minimum pass rate for AEB performance

evaluation? For example, a vehicle is considered to meet the AEB performance if it

passes two-thirds of the 20 unique combinations of test conditions and test speeds (i.e.,

passes 14 unique combinations of test conditions and test speeds).

(43) As fused camera-radar forward-looking sensors are becoming more prevalent in the

vehicle fleet, and the Agency has not observed any instances of false positive test

failures during any of its CIB or DBS testing, is it appropriate to remove the false

positive STP assessments from NCAP’s AEB (i.e., CIB and DBS) evaluation matrix in

this NCAP update? Why or why not?

(44) For vehicles with regenerative braking that have setting options, the Agency is

proposing to choose the “off” setting, or the setting that provides the lowest

deceleration when the accelerator is fully released. As mentioned, this proposal also

applies to the Agency’s PAEB tests. Are the proposed settings appropriate? Why or

why not? Will regenerative braking introduce additional complications for the

Agency’s AEB and PAEB testing, and how could the Agency best address them?

(45) Should NCAP adopt any additional AEB tests or alter its current tests to address the

“changing” rear-end crash problem? If so, what tests should be added, or how should

current tests be modified?

130

(46) Are there any aspects of NCAP’s current FCW, CIB, and/or DBS test procedure(s) that

need further refinement or clarification? If so, what refinements or clarifications are

necessary, and why?

3. FCW and AEB Comments Received in Response to 2015 RFC Notice

NHTSA received several comments in response to the December 2015 notice pertaining

to NCAP’s DBS and CIB tests. These included comments on FCW effective time-to-collision

(TTC), false positive test scenarios, procedure clarifications, expanding testing, and the AEB

strikeable target. These will be discussed over the next few sub-sections.

a. Forward Collision Warning (FCW) Effective Time-to-Collision (TTC)

In its response to NCAP’s December 2015 notice, BMW suggested that the Agency adopt

an “effective TTC” for NCAP’s FCW test that differs from the “absolute TTC” currently

stipulated in the associated test procedure. The manufacturer contended that the deceleration due

to an activated AEB system effectively prolongs the reaction time for the driver such that “an

FCW warning with AEB intervention at an absolute TTC of 2.0 seconds is assumed to show an

equal or greater effectiveness in comparison to an FCW warning at 2.4 seconds without AEB

intervention.” BMW suggested that if AEB functionality is intrinsic to the frontal crash

prevention system, the assessment of the warning TTC in the FCW performance test should

consider the time gained by AEB deceleration and therefore the Agency should assess the

“effective TTC,” not an “absolute TTC.”

The Agency agrees with BMW that FCW and AEB are interrelated and is thus proposing

to assess the presence of an FCW alert as an integral component of the CIB test. To assess the

adequacy of the FCW alert in that context, the Agency has proposed to evaluate the presence of a

vehicle’s FCW system during its CIB tests by requiring the SV accelerator pedal be fully

released within 500 ms after the FCW alert is issued. If no FCW alert is issued during a CIB

test, the SV accelerator pedal will be fully released within 500 ms after the onset of CIB system

131

braking. If no FCW alert is issued and the vehicle’s CIB system does not offer any braking,

release of the SV accelerator pedal will not be required prior to impact with the POV. The

Agency believes that this proposal is philosophically aligned with BMW’s request, as it would

no longer require the direct assessment of FCW timing relative to an “absolute TTC.” Rather,

FCW timing, and how it relates to the intended onset of CIB activation, would be at the

discretion of the vehicle manufacturer (who will have explicit knowledge of how the operation of

their vehicles’ CIB systems affect the “effective TTC”). That said, the Agency continues to

believe that well-designed FCW alerts can provide significant safety benefits in crash-imminent

rear-end crash scenarios, and encourages vehicle manufactures to present them such that the

driver may be able to respond with sufficient time to avoid a crash (i.e., not to solely rely on CIB

activation for crash avoidance). If a vehicle manufacturer chooses to issue an FCW alert in a

way that assumes a CIB intervention will effectively extend the precrash timeline, but then the

AEB system does not activate under real-world driving conditions, or activates late, drivers may

not have enough time to react to avoid an impending crash.

b. False Positive Test Scenarios

Citing the potential for redundancy with the three active/supplemental braking scenarios

for systems exhibiting lower deceleration rates, Mobileye suggested that the Agency impose a

maximum speed reduction of 2 kph (1.24 mph) for the CIB and DBS tests, or a maximum

duration of braking over the maximum allowable deceleration threshold for the false positive

tests. The STP test is designed to provide an indication as to whether a vehicle’s AEB system

may have a false activation problem. Some vehicles use haptic braking and/or low-level braking

as part of their FCW alert strategy. These brake activations are not intended to slow the vehicle

significantly; rather, they attempt to get the driver’s attention so that he/she will respond to the

crash-imminent situation. That said, it is quite possible that FCW-based braking could reduce

speed more than the 2 kph (1.24 mph) threshold suggested by Mobileye.

132

Recognizing the potential problem for a vehicle to fail the CIB false positive test as a

consequence of how its FCW system was designed to work, NHTSA built some flexibility into

the assessment criteria used to evaluate how the subject vehicle (SV) responds to the STP. In the

CIB test, activations can produce peak decelerations of up to 0.5g, which was beyond any FCW-

based level at the time. In the DBS test, the peak deceleration of a given test trial must not

exceed 150 percent of the average peak deceleration calculated for the baseline test series

performed at the same nominal SV speed. These provisions are intended to tolerate small levels

of deceleration, but not the larger magnitudes indicative of an AEB intervention.

BMW objected to the inclusion of the false positive test scenario in general for both DBS

and CIB systems and raised concerns that such tests “can incentivize vehicle manufacturers to

focus on one artificial situation, instead of considering the myriad of potential real-world traffic

situations.” The manufacturer suggested that if this test scenario remains for DBS, then the

Agency should allow manufacturers to specify a brake pedal application rate limit beyond 279

mm/s (11 in./s) and up to 400 mm/s (16 in./s) for the false positive test scenario, to harmonize

with Euro NCAP requirements. BMW further stated that limiting the rate to 279 mm/s (11 in./s)

could increase a DBS system’s sensitivity, and thereby increase the likelihood of additional false

activation events in the real world. The manufacturer mentioned that as more frontal crash

prevention systems combine both FCW and AEB functionalities, speed should reduce for all

pedal application speeds.

Regarding BMW’s objection to continuing with the false positive test scenario for CIB

and DBS in NCAP, NHTSA notes that it has requested comment on whether eliminating the

false positive tests would be appropriate at this time. As discussed previously, the Agency has

not observed false positive test failures in CIB or DBS testing since these ADAS technologies

were added to NCAP.

133

If NHTSA decides it is appropriate to keep the false positive test scenario for DBS,

BMW requested that manufacturers should be permitted to specify a brake pedal application rate

up to 400 mm/s (16 in./s) since this is the upper brake application rate limit established by Euro

NCAP. In its November 2015 final decision notice for AEB, NHTSA addressed a similar

request from the Alliance, which suggested that the Agency harmonize with Euro NCAP’s brake

application rate range of 200 to 400 mm/s (8 to 16 in./s).

190

At the time, the Agency stated that it

would retain its proposed brake application rate of 254 ± 25.4 mm/s (10 ± 1 in./s) in the DBS

system performance test. In justifying this decision, NHTSA contended that the current

application rate value is well within the range of the Euro NCAP specification. Also, NHTSA

reasoned that the current application rate appears to be a feasible representation of the activation

of DBS systems. DBS systems are designed to stop rather than slow down, but not too fast like

conventional brake assist systems, which typically address emergency panic stop situations

where the brake application rate exceeds 360 mm/s (14.2 in./s). For NHTSA to focus on

evaluating system performance for DBS technology (not conventional brake technology), the

Agency plans to retain the current brake pedal application rate of 254 ± 25.4 mm/s (10 ± 1 in./s)

for the DBS test.

c. Procedure Clarifications

In response to the November 2015 final decision notice, Mobileye asked NHTSA to

clarify the process of releasing the accelerator pedal within 500 ms of the FCW alert prior to

braking. The commenter questioned whether the throttle was gradually released over 500 ms, or

abruptly released over 50 ms. Mobileye also asked that the Agency clarify how braking is

affected if there is no FCW alert, or if the FCW alert occurs very close to the brake activation.

NHTSA notes that the throttle pedal release rate is not restricted in NCAP’s CIB test

procedure. The test procedure requires only that the SV throttle be fully released within 500 ms

190

80 FR 68608 (Nov. 5, 2015).

134

after the FCW alert is issued. As previously mentioned, as part of the Agency’s proposed

changes to the CIB tests, it also intends to include test procedure language stating that if no FCW

alert is issued during a CIB test, the SV accelerator pedal will be released within 500 ms after the

onset of CIB system braking, and that if no FCW alert is issued and the vehicle’s CIB system

does not offer any braking, release of the SV accelerator pedal will not be required prior to

impact with the POV.

With respect to how SV braking is affected, if there is no FCW alert, or if the alert

happens very close to brake activation, different steps are taken for the crash imminent braking

(CIB) and dynamic brake support (DBS) tests.

In the existing DBS tests, the test procedure states that the accelerator pedal must be

released within 500 ms after the FCW alert is issued, but prior to the onset of the manual SV

brake application by a robotic brake controller. The Agency recognizes that this can create an

issue if no FCW alert occurs because the throttle may still be depressed (since no warning was

issued) while the SV brakes are applied by the robot at the prescribed TTC. The Agency has

documented this possibility where the SV throttle and brake pedals are applied at the same time

and provided a recommendation that up to a 250 ms overlap be allowed.

191

In other words, once

the SV driver detects that the robot has applied the brakes, the driver will have 250 ms to release

the accelerator fully. The test would not be valid unless this criterion is met.

Although the Agency has proposed to revise when the manual (robotic) brake application

is initiated to a time that corresponds to 1.0 second after the FCW alert is issued (regardless of

whether a CIB activation occurs after the FCW alert but before initiation of the manual brake

application) if it continues to perform DBS testing in NCAP, it has also requested comment on

191

Forkenbrock, G. J., & Snyder, A. S. (2015, June), NHTSA’s 2014 automatic emergency braking test track

evaluations (Report No. DOT HS 812 166), Washington, DC: National Highway Traffic Safety Administration.

135

appropriate TTCs for the modified DBS test conditions as an alternative to this proposal.

Therefore, NHTSA is also requesting comment on the following:

(47) Would a 250 ms overlap of SV throttle and brake pedal application be acceptable in

instances where no FCW alert has been issued by the prescribed TTC in a DBS test, or

where the FCW alert occurs very close to the brake activation. If a 250 ms overlap is

not acceptable, what overlap would be acceptable?

d. Expand Testing

Magna suggested that NHTSA expand testing to encompass low light and inclement

weather situations. The Agency’s proposal for PAEB systems includes testing under less-than-

ideal environmental conditions (specifically at nighttime). The Agency notes that approximately

half (51 percent) of fatalities caused by rear-end crashes and most MAIS 1-5 injuries (80

percent) occurred under daylight conditions. Furthermore, nearly all fatalities (92 percent) and

injuries (88 percent) stemming from rear-end collisions occurred in clear weather.

192

Having

said that, IIHS’s review of 2009-2016 rear-end crash data suggested that AEB-equipped vehicles

are over-represented for crashes occurring in certain weather conditions, such as snow and ice.

193

Therefore, NHTSA is requesting comment on the following:

(48) Should the Agency pursue research in the future to assess AEB system performance

under less than ideal environmental conditions? If so, what environmental conditions

would be appropriate?

e. AEB Strikeable Target

Numerous commenters recommended that NHTSA harmonize its Strikeable Surrogate

Vehicle (SSV) with the test target used by other testing organizations such as IIHS and Euro

192

Swanson, E., Foderaro, F., Yanagisawa, M., Najm, W. G., & Azeredo, P. (2019, August), Statistics of light-

vehicle pre-crash scenarios based on 2011-2015 national crash data (Report No. DOT HS 812 745), Washington,

DC: National Highway Traffic Safety Administration.

193

Cicchino, J. B. & Zuby, D. S. (2019, August), Characteristics of rear-end crashes involving passenger vehicles

with automatic emergency braking, Traffic Injury Prevention. 2019, VOL. 20, NO. S1, S112–S118,

https://doi.org/10.1080/15389588.2019.1576172.

136

NCAP. The commenters reasoned that harmonization would further advance the implementation

of AEB technology by reducing the development and testing burden and thereby result in lower-

cost systems. Mercedes recommended that NHTSA recognize other targets as being equivalent

devices to the SSV and requested that NHTSA allow vehicle manufacturers the option to choose

which target is used for testing.

Currently, NHTSA uses the SSV as the principal other vehicle (POV) in NCAP testing of

DBS and CIB systems. The SSV is a target vehicle modeled after a small hatchback car and

fabricated from light-weight composite materials including carbon fiber and Kevlar®.

194

Using

this target imposes certain limitations, most importantly the maximum speed it can be operated

at, or be struck by, the SV. Due to its material properties, the SSV can inflict damage to vehicles

that impact it at higher speeds.

Another target, the Global Vehicle Target (GVT), which was referenced earlier with

respect to BSI (blind spot intervention) testing, resembles a white hatchback passenger car. This

three-dimensional surrogate is currently used by other consumer organizations, including Euro

NCAP. It is also used by many vehicle manufacturers in their internal testing to NCAP test

specifications, and by NHTSA to facilitate ADAS research using pre-crash scenarios beyond

those included in the Agency’s FCW, CIB, and DBS test procedures.

195

The GVT consists of 39 vinyl-covered foam pieces (held together with hook and loop

fasteners) that form the structure the outer skins are attached to. It is secured to the top of a

Low-Profile Robotic Vehicle (LPRV) using hook and loop fasteners, which separate upon an

SV-to-GVT collision. When the GVT is hit at low speed, it is typically pushed off the LPRV but

remains assembled. At higher impact speeds, the GVT breaks apart as the SV essentially drives

through it, and can then be reassembled on top of the LPRV.

194

80 FR 68604 (Nov. 5, 2015).

195

Currently, manufacturers use test results from their internal testing and submit them to NHTSA for NCAP’s

recommendation of vehicles that pass its performance testing requirements.

137

The use of this surrogate vehicle would allow the Agency to perform tests at higher

speeds, thus increasing safety benefits. For this reason, the Agency used the GVT in its

characterization study for CIB testing at higher speeds. The SSV initially limited the test speeds

the Agency could adopt for CIB and DBS testing because of concerns over potential damage to

the testing equipment and test vehicle. Using the GVT significantly reduces that possibility for

the test speeds proposed. Also, as future upgrades for NCAP are planned, the GVT can be used

to evaluate more challenging crash scenarios, such as those required for other ADAS

technologies (Intersection Safety Assist and Opposing Traffic Safety Assist). NHTSA has

recently docketed draft research test procedures for these technologies.

196, 197

If, in the future, the

Agency was to consider adopting other test procedures requiring a strikeable target,

incorporating the GVT would allow harmonization across the program.

NHTSA has conducted vehicle testing to evaluate the FCW alert and CIB intervention

onset timing observed using the GVT Revision E and compared that with the timing recorded for

identical tests performed with NHTSA’s SSV benchmark.

198

Three light vehicles and three rear-

end crash scenarios were used for this evaluation. A secondary objective of this study was to

assess the characteristics and durability of the GVT for various test track configurations,

specifically its dynamic stability and in-the-field reconstruction time after being struck by a test

vehicle. GVT stability was evaluated using straight line and curved path maneuvers at various

speeds and lateral accelerations. Reconstruction times of the GVT after impact were examined

using different impact speeds, directions of impact, and assembly crew sizes.

196

National Highway Traffic Safety Administration (2019, September), Intersection safety assist system

confirmation test: Working draft, http://www.regulations.gov, Docket No. NHTSA-2019-0102-0006.

197

National Highway Traffic Safety Administration (2019, September), Opposing traffic safety assist system

confirmation test: Working draft. http://www.regulations.gov, Docket No. NHTSA-2019-0102-0008.

198

Snyder, A.C., Forkenbrock, G.J., Davis, I.J., O’Harra, B.C., & Schnelle, S.C. (2019, July), A test track

comparison of the global vehicle target and NHTSA’s strikeable surrogate vehicle (Report No. DOT HS 812 698),

Washington, DC: National Highway Traffic Safety Administration.

138

Overall, the results from the study suggested that the onset timing of FCW and CIB

systems observed during rear-end tests performed with the GVT was similar to that recorded for

the SSV.

199

The GVT was also found to be physically stable and remained affixed to the robotic

platform used to facilitate its movement during the high-speed longitudinal tests as well as those

performed at the limit of the platform’s lateral road holding capacity. Although the time between

test trials was longer than that associated with use of the SSV, GVT reassembly tests

demonstrated that the GVT could be reconstructed in a reasonable time between tests after being

struck. However, the physical reconstruction time is one of three considerations when

determining the time between tests when the GVT is used. After being reassembled and secured

to the top of the robotic platform, the platform must re-establish its communication with the

other equipment needed to perform the tests, and a “zero-offset” check is used. This check not

only ensures the GVT orientation relative to the platform remains consistent for all tests, but also

confirms the distance from the SV to the GVT at the point of impact is accurately reported as

zero when the two first make contact.

NHTSA proposes to use the GVT in lieu of the SSV in future NCAP testing. Similar to

that noted earlier regarding the use of the articulated pedestrian mannequins, the use of the GVT

provides another opportunity for NHTSA to harmonize with other consumer information safety

rating programs as mandated by the Bipartisan Infrastructure Law. Comments are sought on its

adoption regardless of whether modifications are made to test speeds, deceleration, test

scenarios, combining test procedures, et cetera, as has been discussed.

The Agency also recognizes that there have been ongoing revisions to the GVT to

address its performance in other crash modes that exercise different ADAS applications. At this

time, NHTSA believes the latest Revision G is appropriate for testing in NCAP. However, for

199

Comparable observations were made upon review of test data from the Agency’s CIB characterization testing.

Upon review of test data from the Agency’s CIB characterization testing, FCW and CIB onset timings for identical

vehicles were highly comparable regardless of whether the SSV or GVT Revision G targets were used.

139

the purpose of AEB testing only, NHTSA is proposing to accept manufacturer verification data

for AEB tests conducted using GVT Revision F.

200,201

It is the Agency’s understanding that

Revision G incorporates changes to the front, side, and oblique aspects of Revision F.

202

NHTSA believes that modifications implemented for Revision G have not altered the physical

characteristics of the rear of the target such that a vehicle’s performance in the rear-end crash

mode would be impacted. The Agency requests comment on:

(49) the use of the GVT in lieu of the SSV in future AEB NCAP testing,

(50) whether Revisions F and G should be considered equivalent for AEB testing, and

(51) whether NHTSA should adopt a revision of the GVT other than Revision G for use in

AEB testing in NCAP.

With respect to Mercedes’ request that NHTSA consider several targets and allow

manufacturers the option to choose which target is used for testing, the Agency does not believe

such an approach is feasible. The Agency currently accepts and uses, for recommendation

purposes on www.NHTSA.gov, data submitted by vehicle manufacturers for internal CIB and

DBS testing that was conducted using a target other than the SSV, such as the Allgemeiner

Deutscher Automobil-Club e.V (ADAC) target, which was previously used by Euro NCAP and

200

While the Agency used GVT Revision E in its comparative testing with the SSV, and it believes that no

significant differences exist between Revision E and Revision F that would affect AEB test results, the Agency does

not believe it is necessary to accept from vehicle manufacturers AEB test data that was derived using Revision E

because Revision E is no longer in production. Therefore, the Agency believes that any OEM data that is submitted

should reflect the use of GVT Revision F or Revision G.

201

Although the Agency used GVT Revision E in its comparative testing with the SSV, the Agency does not believe

that modifications made for Revision F would have changed the results of that testing. It is the Agency’s

understanding that several modifications were made to the rear of Revision E, which included adding additional

radar material to the bottom skirt of the target to attenuate internal reflections, and reducing the slope of the rear top

portion of the hatchback to increase the power of the radar return.

202

To improve the real-world characteristics from the front and side of the target, several changes to the radar

treatment were integrated into the components of the GVT body for Revision G compared to Revision F, including

changes to the skin and wheel treatment. There were also some minor shape changes to the front of the GVT body to

improve front radar return and to the side to improve the ability to hold its shape.

http://www.dynres.com/2020/02/25/the-new-global-vehicle-target-gvt-has-arrived/.

140

IIHS.

203

However, during its system performance verification testing, the Agency has observed

several test failures, which may be attributed to differences in target designs.

In NHTSA’s November 2015 AEB final decision notice,

204

NHTSA stated that

manufacturers do not need to use the SSV to generate and submit self-reported test data in

support of their AEB systems that pass NCAP’s system performance requirements and are

recommended to consumers on the Agency’s website. However, if the vehicle does not pass

NCAP’s system performance criteria for AEB systems during the program’s random system

performance verification testing, the Agency would remove the recommendation from its

website. To uphold the credibility of the program and reasonably assure that consumers are

receiving vehicles that meet a specified minimum performance threshold, NHTSA believes that

it is critical to accept self-reported data from manufacturers that was obtained using tests

conducted in accordance with NHTSA test procedures. As such, NHTSA is proposing not to

accept vehicle manufacturer test data that was derived from an alternative test target other than

that which is specified in NCAP’s test procedures.

IV. ADAS Rating System

NHTSA is planning to create a rating system based on assessments related to the

performance of ADAS technologies, including, but not necessarily limited to, the technologies

already part of the program and others proposed above. Currently, NCAP places a check mark

by the relevant ADAS technology on NHTSA’s website, www.nhtsa.gov, if two conditions are

met: (1) a vehicle is equipped with the safety technology recommended by NHTSA; and (2) the

system meets NCAP’s performance specifications. Consumers are encouraged to look for

vehicles equipped with ADAS that meet NCAP’s performance tests, which are intended to

203

80 FR 68604 (Nov. 5, 2015).

204

80 FR 68607 (Nov. 5, 2015).

141

establish a minimum level of performance on which consumers can rely and compare among

vehicles equipped with similar technologies.

In the Agency’s December 2015 notice, NHTSA discussed a series of point values for the

ADAS technologies at that time. These points would have been used in a star rating system for

these technologies. Vehicles with ADAS that met the criteria set forth in the Agency’s test

procedures would earn full points if offered as standard equipment on a particular model and half

points if offered only as optional equipment for that model. In response to that proposal,

commenters provided mixed support regarding the feasibility and appropriateness of developing

such an ADAS rating system versus the current process of just identifying the presence of

recommended technologies with check marks.

205

Proponents of a rating system were generally

supportive of the broad concept of rating ADAS, but did not propose specific suggestions for

how the Agency could develop such a rating system. Some commenters responded that ADAS

technologies have not yet matured to the point that a rating system would be appropriate, while

others believed that one could be developed. In the responses for the October 1, 2018 public

meeting, support still varied, even when the discussion was more focused on how the FAST Act

mandate to provide crash avoidance information on the Monroney label might be fulfilled in the

context of an ADAS rating system.

A. Communicating ADAS Ratings to Consumers

As mentioned previously, NHTSA’s current method of providing ADAS information to

consumers conveys which systems meet NCAP’s system performance requirements, but

provides no overall ADAS technology rating for the vehicle. However, as more emerging

ADAS technologies are available in the market, the Agency believes that a rating mechanism for

these systems would be more beneficial for consumers because it could better distinguish the

technologies, including different levels of system performance and the technologies’ life-saving

205

https://www.regulations.gov, Docket No. NHTSA-2015-0119.

142

potential, rather than simply listing how many technologies a given vehicle is equipped with that

meet NCAP’s system performance requirements. As will be discussed in the sections that

follow, ADAS ratings could be communicated to consumers using stars, medals, points, or other

means, thereby allowing them to make better-informed decisions. Also, the ratings could be

based on the safety benefit potential afforded by vehicles’ ADAS technologies and system

performance. In addition, NHTSA plans to explore several approaches on how to present such

rating information in the Agency’s planned consumer research. In this RFC, NHTSA is

soliciting input solely on the creation of an ADAS rating system, not the visual representation or

placement of that rating system at points of sale. As described in greater detail below, issues

related to the visual representation and placement of the rating system at points of sale will be a

topic covered in future notices and research.

1. Star Rating System

NCAP currently uses 1 to 5 stars to communicate vehicle crashworthiness ratings to

consumers, with both ratings for the individual tests and an overall rating. Given the familiarity

that consumers have with NHTSA’s current 5-star ratings system, the Agency could also

consider the use of stars for a future ADAS rating system. However, the Agency has some

reservations about pursuing such an approach.

A future star-based ADAS rating system could produce lower ratings for technologies

than consumers are accustomed to seeing in crashworthiness and rollover resistance tests, and

may cause unnecessary consumer confusion about the additional safety the technology on their

vehicle provides. For instance, although NHTSA believes ADAS could potentially add

significant safety benefits in addition to the crashworthiness protection afforded on vehicles, the

Agency questions whether consumers would interpret 1- and 2-star ADAS ratings as conveying

added benefits beyond the crashworthiness protection offered by a vehicle. In addition, vehicles

that do not have any ADAS ratings could mistakenly be interpreted to have an advantage (i.e.,

143

additional safety benefits) over those that have low ADAS star ratings. Thus, vehicles that have

low ADAS star ratings could inadvertently discourage consumers from considering ADAS in

their purchasing decisions, when in fact, those vehicles with 1- and 2-stars may offer significant

safety benefits over their unrated peers.

Given these concerns, the Agency could consider reserving star ratings to convey

crashworthiness results only and distinguish ADAS ratings by using another visualization

approach, such as a medals system or points-based system.

2. Medals Rating System

Another potential method of presenting ADAS rating information to consumers could be

a three-tiered award system similar in concept to Olympic medals. Presumably, most consumers

are already familiar with the designations of bronze, silver, and gold as increasingly more

prestigious levels of achievement.

Using an awards system (e.g., medals) rather than stars to represent NCAP’s rating of

ADAS technologies would not only distinguish ADAS grades from crashworthiness ratings, but

also visually communicate that the two ratings are conveying different types of vehicle safety

information. However, it could cause consumer confusion by having two separate rating systems

that consumers would need to consider and, to the extent there is a divergence between the two

systems, potentially weigh against one another for a given vehicle.

3. Points-Based Rating System

NHTSA could use points to convey ADAS rating information. Points could be used in

lieu of stars or medals or in addition to these alternative rating communication concepts, and they

may serve as the basis for any of the potential rating system approaches discussed in the sections

that follow. One advantage of a points-based system is that it can provide improved delineation

in ratings, thus benefiting consumers who may want to compare ratings between several vehicle

models. However, the inherent granularity of a points-based system may cause consumer

144

confusion if conveyed in addition to another, coarser, communication rating concept, such as

stars or medals. As mentioned previously, NHTSA plans to conduct consumer research

surrounding the concept of an overall NCAP rating that would combine results from

crashworthiness, rollover resistance, and ADAS technology testing.

4. Incorporating Baseline Risk

Another consideration for the Agency that may add value to an ADAS rating system is

the notion of conveying a vehicle’s performance relative to the baseline (or average)

performance observed for today’s vehicle fleet. As detailed later in this notice, this concept is

currently an element of NCAP’s crashworthiness rating system. Star ratings generated in NCAP

today are a measure of how much more (or less) occupant protection (in terms of injury risk) a

given vehicle affords when compared to an “average” vehicle. The Agency could consider

incorporating the baseline concept when developing an ADAS rating system as well. For

instance, today’s “average” vehicle may achieve 60 out of a possible 100 points (or 3 out of 5

stars) during NCAP’s testing. This score (or rating) may translate to a 30 percent reduction in

the risk of crashes, injuries, deaths, etc. Scores (or ratings) for future vehicles, which could also

potentially be tied to a percent reduction in crashes, could be compared relative to the baseline

rating of today’s fleet, thus affording consumers the opportunity to compare scores (or ratings)

for vehicles spanning different model years.

B. ADAS Rating System Concepts

Just as there are several ways to communicate ADAS ratings to consumers, there are also

several ways to rate ADAS technologies, a few of which are discussed below. As each of these

rating system concepts center around vehicle performance in NCAP tests, it was necessary to

consider the primary components of these tests during concept development.

1. ADAS Test Procedure Structure and Nomenclature

145

As discussed extensively in this notice, each ADAS technology and associated test

procedure the Agency is considering for inclusion in NCAP has the potential to address a real-

world safety problem. Each test procedure is designed to replicate certain injurious and fatal

real-world events (termed “scenarios” in this new rating concept) that can be approximated in a

laboratory setting to assess the capabilities of a given ADAS. Within each scenario, the Agency

defines test conditions to replicate types of real-world incidents. Within each test condition, one

or more test variants (as illustrated in Figures 1 and 2 below) that assess the limitations of each

ADAS technology under that test condition is also defined.

206

Finally, for each test variant, the

technology would have to pass a certain number of trials to receive credit for that part of the

ADAS rating. Figure 1 illustrates a generic structure for describing a given ADAS test

procedure and its nomenclature in NCAP.

Figure 1: Generic ADAS test procedure nomenclature

206

In certain test conditions that do not have a multitude of assessments (e.g., test condition variants), the test

condition and assessment would be one and the same.

Real-world pre-

crash scenarios

Test procedure

Scenario 1

Test Condition

Variant 1

Trial 1

Trial 2

Variant 2

Trial 1

Trial 2

Test Condition

Trial 1

Trial 2

Scenario 2

Test Condition

Variant 1

Trial 1

Variant 2

Trial 1

146

The above methodology and diagram can be illustrated further using one of the ADAS

technologies discussed in this document, PAEB. PAEB is intended to address a real-world

safety issue involving vulnerable road users, like pedestrians. The current test procedure is

designed to replicate S1 and S4 scenarios (vehicle heading straight with a pedestrian crossing the

road, and a vehicle heading straight with a pedestrian walking along or against traffic,

respectively). Within each scenario, one or more test conditions are defined. For example,

within the S1b test scenario (as previously discussed), several test condition variants are defined.

In this case, the same test condition would have to be executed at various speeds (test condition

variants). Finally, NHTSA would prescribe the number of trials for which the system would

have to exhibit conformance to receive credit for these particular test condition variants and, in

turn, scenario. Figure 2 illustrates this example.

147

Figure 2: Scenario Sb1 of the proposed NCAP PAEB test procedure

To illustrate further the multitude of assessments simplified in Figure 1, certain test

scenarios only include one test condition and one test variant. A specific example of this would

be the previously mentioned Lead Vehicle Stopped (LVS) scenario, evaluated as part of the

Crash Imminent Braking (CIB) test procedure, where the Subject Vehicle (SV) encounters a

stopped Principal Other Vehicle (POV) on a straight road moving at 40.2 kph (25 mph). This

example is illustrated in Figure 3.

Scenario •S1

Test Condition

• b (crossing adult pedestrian walking nearside,

50% overlap)

Var i ants

• 10, 20, 30, 40,

50, 60 kph (6.2,

12.4, 18.6, 24.9,

31.1, 37.3 mph)

Trials

• 1-5 trials

148

Figure 3: LVS Scenario of the NCAP CIB test procedure

2. Percentage of Test Conditions to Meet – Concept 1

Given the test procedures’ structure, an ADAS rating system could be designed with

standards of increasing stringency that must be achieved to receive higher award levels (as

shown in Table 7 below). In such a system, different ADAS technologies, each with a related

test procedure (e.g., FCW, CIB, LKS), are combined into categories where each technology

addresses a similar crash problem. For instance, ADAS Category 1 in Table 7 could represent

the Forward Collision Prevention category that would be comprised of the three forward

collision prevention technologies, FCW, CIB, and DBS. Vehicles would have to meet increasing

numbers of test conditions across all test procedures in that particular ADAS category (i.e., three

test procedures for the example given) to achieve higher ratings (e.g., medals, stars, points). For

the example rating system concept shown in Table 7, 50 percent of test conditions would have to

be met to achieve a bronze award, 75 percent to achieve a silver award, and 100 percent to

Scenario •LVS

Test Condition

• SV encounters stopped

POV on straight road. SV

at 40 kph (24.9 mph),

POV stationary (0

kph/mph)

Var i ant

• Inherent to

test

condition

Trials

• 1-5 trials

149

achieve a gold award for each ADAS category.

207

The lowest ADAS rating among the

categories could serve as the overall ADAS award if a summary rating is established across all

included ADAS technologies. Alternatively, an overall ADAS award could reflect the average

ADAS rating amongst the technology categories.

Table 7: 3-Tier ADAS Rating System Concept 1

All Test Procedures & Conditions in ADAS

Category

Award

Bronze Silver Gold

50% of Test

Conditions

75% of Test

Conditions Met

100% of Test

Conditions

ADAS Category 1 Meets Did not meet Did not run Bronze

ADAS Category 2 Meets Meets Meets Gold

ADAS Category 3 Meets Did not meet Did not run Bronze

ADAS Category 4 Meets Meets Did not meet Silver

Overall ADAS Award Bronze

3. Select Test Conditions to Meet – Concept 2

Table 8 demonstrates another possible NCAP ADAS rating system concept. As with

Concept 1, ADAS technologies are grouped into categories that address similar crash problems.

Instead of having to meet a percentage of all test conditions, NCAP could specifically require

certain test conditions to be met for each of three award levels. These award levels could be

based on the following increasingly challenging delineations:

(1) Bronze (Basic performers) – test conditions that are achievable for current systems to

meet;

(2) Silver (Advanced performers) – test conditions that are more difficult for current systems

to meet but are more easily achievable than the current known system limitations; and

207

When ‘Did not meet’ is listed for an ADAS category, the vehicle failed to pass the requirements for the test

condition/variant when tested. ‘Did not run’ may be used to signify that the vehicle is not equipped with the

technology to pass the related test procedure(s), and as such, the tests were not conducted.

150

(3) Gold (Highest performers) – test conditions that approach the current limits of system

testing feasibility, vehicle operations, and event extremes.

Depending on a given technology’s test procedure, the number of test conditions, test

condition variants, and trial passes necessary to meet the Agency’s requirements could vary.

Thus, the ADAS performance requirements necessary for reaching each subsequent award level

could be based on meeting a single test condition variant or meeting a number of test conditions.

To explain further in the context of Table 8, ADAS Group 1 could be the Lane Keeping

Assistance (LKA) technology category, where technology 1 could be LDW, and technology 2

could be LKS. In this example, the vehicle’s LDW system meets all applicable test conditions

(bronze, silver, gold). However, its LKS system fails to meet the test conditions required for

silver, but meets the test conditions to earn bronze. Therefore, the highest award this vehicle

could achieve for the LKA category would be bronze, as it is the highest award achieved by both

of the technologies (LDW and LKS) included in the LKA category. Similar to Concept 1, the

lowest or average ADAS rating amongst the category groups could serve as the overall ADAS

award if a summary rating is established across all included ADAS technologies.

Table 8: 3-Tier ADAS Rating System Concept 2

Bronze Test Conditions Silver Test Conditions

Gold Test

Conditions

ADAS

Group

Award

ADAS Group 1 1 2 3 1 2 1

Bronze Tech 1 Meets Meets Meets Meets Meets

Tech 2 Meets Meets Meets Meets Did not meet Did not ru

ADAS Group 2 1 2 3 1 2 1

Gold Tech 1 Meets Meets Meets Meets Meets Meets

Tech 2 Meets Meets Meets Meets

ADAS Group 3 1 2 3 1 2 1

Bronze

Tech 1 Meets Meets Meets Did not meet Did not ru

Did not ru

ADAS Group 4 1 2 3 1 2 1

Silver

Tech 1 Meets Meets Meets Meets Meets Did not meet

Overall ADAS

Award

Bronze

151

A more detailed example of this ADAS rating system concept, which uses some of the

test conditions and test condition variants discussed in this document (distinguished by variables

such as speed), is shown below in Table 9. In this example, check marks are used to indicate that

the vehicle’s ADAS technology has met the requirements for a given test procedure’s conditions

and test condition variants. An “X” symbol is used to indicate where vehicles did not meet the

test condition and/or variants, either because the vehicle was not equipped with the technology

and therefore could not be tested, or because the vehicle’s technology was tested, but failed to

meet the test procedure requirements. Units are in kph unless otherwise noted.

To further explain the three-tier rating system illustrated in Table 9 with context, ADAS

Group 3 in the example utilizes Blind Spot Detection (BSD) to demonstrate multiple test

conditions and test condition variants. BSW (categorized as Technology 1 for the BSD

grouping) has five test condition variants, and BSI (categorized as Technology 2 for the BSD

grouping) includes three test condition variants. In order for BSD to achieve a bronze award in

this example, the BSW system must meet the three test condition variants included for this

technology under the ‘Bronze Test Conditions/Variants’ heading. No BSI test conditions, or test

condition variants, must be met. In order for BSD to achieve a silver award, BSW must meet

two test conditions (comprised of five test condition variants) and BSI must meet two test

conditions, both of which are included under the ‘Silver Test Conditions/Variants’ heading. If

the vehicle was also able to meet the third test condition included in the BSI test procedure, ‘SV

Lane Change w/ Closing Headway 72.4/80.5’, which is included under the ‘Gold Test

Conditions/Variants’ heading in Table 9, the vehicle would earn a gold award. In the Table 9

example, however, BSI does not meet one of the silver test conditions/variants (‘SV Lane

Change w/ Constant Headway 72.4/72.4’). Consequently, in this example, BSD achieves the

next lowest award—bronze.

152

Table 9: Example of 3-Tier ADAS Rating System Concept 2

ADAS

Group

Bronze Test Conditions/Variants Silver Test Conditions/Variants Gold Test Conditions/Variants

ADAS

Group

Awar

Forward

Collision

Preventio

(ADAS

Grou

1 2 3 1 2 3 4 5 6 1 2 3 4 5 6 7 8

Bronz

FCW

LVS LVD LVM

  

DBS

LVM

40.2/16.1

STP 40.2 STP 72.4 LVD

LVM

72.4/32.2

LVS

   



CIB

LVM

40.2/16.1

STP 40.2 STP 72.4 LVD

LVM

72.4/32.2

LVS

    

Lane

Keeping

Assistanc

(ADAS

Group 2)

1 2 3 1 2 3 4 5 6 1 2 3 4 5 6 7 8

Gold

LDW

Solid

White

Left/Righ

Dashed

Yellow

Left/Rig

Botts’ Dots

Left/Right

  

LKS

Solid

White

Left/Righ

t, 0.2 m/s

Dashed

White

Left/Righ

t, 0.2 m/s

Solid

White

Left/Righ

t, 0.3 m/s

Dashed

White

Left/Righ

t, 0.3 m/s

Solid

White

Left/Righ

t, 0.4 m/s

Dashed

White

Left/Righ

t, 0.4 m/s

Solid

White

Left/Righ

t, 0.5 m/s

Dashed

White

Left/Righ

t, 0.5 m/s

       

Blind

Spot

Detection

(ADAS

Grou

1 2 3 1 2 3 4 5 6 1 2 3 4 5 6 7 8

Bronz

BSW

Converge

Diverge

Left/Righ

Pass-by

72.4/80.5

Left/Rig

Pass-by

72.4/88.5Left/Rig

Pass-by

72.4/96.6

Left/Righ

Pass-by

72.4/104.

Left/Righ

    

BSI

SV Lane

Change

Constant

Headway

72.4/72.4

SV Lane

Change

Constant

Headway

False

Positive

SV Lane

Change

Closing

Headway

72.4/80.5

 

153

Forward

Pedestria

n Impact

Avoidanc

e (ADAS

Grou

1 2 3 1 2 3 4 5 6 1 2 3 4 5 6 7 8

Silver

PAEB

S1f,

40.2/4.8

S1g,

40.2/4.8

S1a,

16.1/4.8

S1b,

16.1/4.8

S1c,

16.1/4.8

S1d,

16.1/4.8

S4a,

16.1/

S4b,

16.1/

S1a,

40.2/4.8

S1b,

40.2/4.8

S1c,

40.2/4.8

S1d,

40.2/4.8

S1e,

40.2/8.

S4a,

40.2/

S4b,

40.2/

S4c,

40.2/4.

         

 



Overall

ADAS

Award

Bronze

154

The approach presented in Tables 8 and 9 would address the Agency’s desire to introduce

a dynamic ADAS rating system. As technologies become more mature, the Agency expects

ADAS system performances will begin to exceed NCAP testing requirements, and as such,

systems will have an easier time meeting the required test conditions across all test procedures.

The Agency could begin providing information on higher performing systems by periodically

increasing the stringency of requirements to achieve the highest NCAP ratings. Lower award

levels could be reserved for test conditions that are easily achieved by ADAS in the current

vehicle fleet. Higher award levels could be reserved for test conditions that current ADAS have

difficulty achieving, or for new test scenarios (e.g., PAEB S2 or S3), conditions (e.g., using a

motorcycle or cyclist as the POV), or variants (e.g., increased SV/POV speeds, decreased

headways, additional weather conditions, varying deceleration rates) that are added to the

program over time. This approach is expected to continue to provide consumers information on

vehicle safety designs that introduce truly exceptional ADAS performance compared to their

peers. It should also incentivize vehicle manufacturers to improve their ADAS capabilities to

meet consumers’ expectations for system performance.

Along these lines, NHTSA could also introduce a slight deviation to rating system

Concept 2. In this deviation, not only would vehicles have to meet the most demanding

requirements across all ADAS test procedures to receive higher ratings, but also the Agency

could set the performance target for the highest level rating (gold, 5 stars, maximum points, etc.)

for those test conditions that are required for an ADAS technology that is just emerging in the

marketplace, such as Intersection Safety Assist (ISA), mentioned later in this notice. In doing so,

consumers could be assured that purchasing a vehicle that earns the highest award level would

offer the most advanced ADAS capabilities available at that time.

4. Weighting Test Conditions Based on Real-World Data – Concept 3

155

The Agency believes it is important to develop an ADAS rating system that is not only

flexible (i.e., one that can adapt or change over time) to keep pace with advancements in

technologies, but also effective in providing consumer information that encourages the

proliferation of life-saving technology. As such, a third rating system concept that the Agency

could consider would be one which weights the technology groups based on the target population

data and effectiveness attributable to each technology to derive the overall ADAS award. In

essence, the more critical, more lifesaving, and/or more advanced/effective technology systems

would have more contribution (i.e., be worth more) in the rating system. Furthermore, for a

given technology group, the Agency could weight the test conditions that approximate more

frequent or injurious real-world events so that they have more influence in the rating for that

group. The selected evaluation method could be normalized in such a way that the results of

each test condition within a scenario could be appropriately combined and concisely presented

for consumer information or ratings purposes. Such an approach could also be incorporated for

either Concept 1 or Concept 2, discussed above.

Utilizing real-world data to inform the structure of a future ADAS rating system is

challenging for several reasons. For one, there is no single metric (such as target crash

populations, fatalities, or injuries) that can be used to weight every technology appropriately in a

rating system when both the related real-world safety problem and meaningful influence are

considered. In an effort to correlate rating system weights directly with potential real-world

safety benefits, too little weight may be assigned to technologies that have lower target

populations (such as those for Blind Spot Detection) compared to technologies that have much

higher target populations (such as those for Forward Collision Prevention). Thus, the Agency is

concerned that it may be possible for manufacturers to offer one or two ADAS systems that

perform well in the NCAP tests, if those technologies with higher target populations are

apportioned significant weight in a rating system, while choosing not to include the other, lower-

156

weighted technologies on their vehicles, or opting to include them even if the systems perform

poorly. Therefore, the Agency believes that it is critical to find an acceptable balance between

weights dictated solely by real-world data and those that ensure each component provides a

meaningful contribution to the rating system. In essence, each technology should be apportioned

within the rating system such that it provides a significant contribution while also reflecting the

relative safety improvement that each technology may afford consumers.

Changes in target population data (based on real-world crashes) and improvements made

to ADAS technologies over time pose additional challenges for the Agency in using real-word

data and system effectiveness estimates to inform appropriate weights or proportions to assign to

the individual test conditions or the corresponding test condition variants in an ADAS rating

system.

208

As technology systems improve to meet NCAP test scenarios/conditions, system

effectiveness estimates may increase. Furthermore, as mentioned earlier in this notice, the real-

world crash data may change as technologies are designed to address certain crash scenarios, but

not others. Ideally, the Agency would adjust rating system weights to keep pace with these

changes, as this would align with NHTSA’s goal of developing a flexible ADAS rating system

that can respond appropriately to improvements or changes seen for the fleet. Unfortunately,

real-world data for system performance advancements is not always readily available to support

dynamic program upgrades, as the crash data, which takes time to reflect changes in the vehicle

fleet accurately, lags system updates and deployments.

Having said that, the Agency sees merit in using available real-world data, specifically

target populations, to determine which ADAS technologies should be considered for inclusion in

the program. The additional time between technology development and NHTSA’s ability to

collect real-world data on target populations has proven in the past to be sufficient to ensure that

208

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

157

the technology is mature prior to considering it in NCAP. As mentioned previously, the four

ADAS technologies discussed in this proposal focus on the most frequently occurring and/or

most severe crash types, which the Agency believes is a feasible and prudent approach to use

when considering whether an ADAS technology should be incorporated into NCAP. NHTSA

will continue to leverage all information and safety studies on ADAS technologies, such as those

cited in this notice, to support the Agency’s proposal. In addition, NHTSA plans to leverage all

available data to assess real-world insights into advanced safety technology performance.

5. Overall Rating

As discussed herein, there are many considerations when developing a potential ADAS

rating system. These include: (1) what type of system to adopt; (2) whether to use points,

medals, or awards to convey ratings; and (3) whether to weight system components based on

real-world data. Another consideration is whether to have an overall rating. Although the

concepts discussed thus far have included an overall rating, NHTSA could also simply list

individual ratings for the included ADAS technologies, but not adopt an overall rating. NHTSA

believes that consumers may have preferences as to which specific ADAS technologies they

would or would not want on their vehicles and may be interested only in how those individual

technologies perform in the Agency’s testing, not in how the vehicle systems perform overall.

The Agency notes that the assignment of ratings for individual technologies could simply

supplement the NCAP program’s existing list approach, or individual technology ratings could

be listed concurrently with an overall rating. Thus, the Agency requests comment on whether an

overall rating system is necessary and, if so, whether it should replace or simply supplement the

existing list approach.

With regard to a future ADAS rating system, the Agency seeks comments on the

following:

(52) the components and development of a full-scale ADAS rating system,

158

(53) the aforementioned approaches as well as others deemed appropriate for the

development of a future ADAS rating system in order to assist the Agency in

developing future proposals,

(54) the appropriateness of using target populations and technology effectiveness estimates

to determine weights or proportions to assign to individual test conditions,

corresponding test combinations, or an overall ADAS award,

(55) the use of a baseline concept to convey ADAS scores/ratings,

(56) how best to translate points/ratings earned during ADAS testing conducted under

NCAP to a reduction in crashes, injuries, deaths, etc., including which real-world data

metric would be most appropriate,

(57) whether an overall rating system is necessary and, if so, whether it should replace or

simply supplement the existing list approach, and

(58) effective communication of ADAS ratings, including the appropriateness of using a

points-based ADAS rating system in lieu of, or in addition to, a star rating system.

In responding to these approaches, or in developing new approaches for consideration,

NHTSA requests that commenters consider a potential ADAS rating system that would allow

flexibilities for continuous improvements to the program and cross-model year comparisons. In

this notice, the Agency is seeking feedback on the appropriateness of the test scenarios, test

conditions, test condition variants, and number of trials within each test variant for the four

proposed technologies (PAEB, LKS, BSW, and BSI) discussed in this RFC, in addition to the

four technologies currently included in NCAP. After NHTSA reviews comments in response to

this notice, particularly those in response to questions raised within each of the ADAS

technology sections and the rating system concepts discussed herein, the Agency anticipates

finalizing the related test procedures and would then develop the selected ADAS rating system

based on the technologies, test scenarios, test conditions, etc. that have support for incorporation

159

into the program. Until NHTSA issues (1) a final decision notice announcing the new ADAS

rating system and (2) a final rule to amend the safety rating section of the vehicle window sticker

(Monroney label), the Agency plans to continue assigning NCAP credit, using check marks on

www.nhtsa.gov, to vehicles that (1) are equipped with its recommended ADAS technologies, and

(2) pass the applicable system performance test requirements.

V. Revising the Monroney Label (Window Sticker)

The third part to this notice relates to the Fixing America’s Surface Transportation

(FAST) Act, which includes a section that requires NHTSA to promulgate a rule to ensure crash

avoidance information is displayed along with crashworthiness information on window stickers

(also known as Monroney labels) placed on motor vehicles by their manufacturers.

209

At the

time of the FAST Act, NHTSA was already in the process of developing an RFC notice to

present many proposed updates to NCAP, including the evaluation of several new ADAS and a

corresponding update of the Monroney label.

NHTSA currently requires vehicle manufacturers to include safety rating information,

obtained from NHTSA under its NCAP program, on the Monroney labels of all new light

vehicles manufactured on or after September 1, 2007 (49 CFR part 575). This requirement was

mandated by Section 10307 of the Safe, Accountable, Flexible, Efficient Transportation Equity

Act; A Legacy for Users (SAFETEA-LU). The purpose of the law is to ensure that vehicle

manufacturers provide consumers with relevant vehicle safety ratings information on all new

light vehicles at the point of sale so that they can make informed purchasing decisions.

Although the safety rating information included on the Monroney label has provided

consumers with valuable information at the point of sale, there are limitations with the current

label for NCAP. For instance, currently the vehicle safety rating section of the Monroney label

only includes vehicle performance information for the crashworthiness program in NCAP

209

Section 24321 of the FAST Act, otherwise known as the “Safety Through Informed Consumers Act of 2015.”

160

(known as the 5-star safety ratings), which is comprised of a full-frontal impact test, a side

impact barrier test, a side impact pole test, a static measurement of the vehicle’s stability factor,

and a dynamic assessment of the vehicle’s risk to rollover in a single-vehicle crash. The other

consumer information program in NCAP, which is the ADAS technologies assessment, is not

included in the current vehicle safety rating section of the Monroney label. This information is

only available on www.nhtsa.gov, along with the 5-star safety ratings information.

210

Thus, NHTSA plans to issue a notice of proposed rulemaking (NPRM) in 2023 to include

ADAS performance information from NCAP in the vehicle safety rating section of the

Monroney label, as mandated by the FAST Act. However, NHTSA seeks a flexible means to

keep pace with the technological advancement and the frequent development of new ADAS

technologies while also providing adequate public participation and transparency. NHTSA

would like to develop a way to allow the Agency both to convey NCAP vehicle safety

information in the safety rating section of the Monroney label and minimize the number of

rulemaking actions needed each time the Agency incorporates a new technology in NCAP.

At this time, NHTSA believes it may be able to achieve these goals by adopting all or

some combination of the following three main categories for the safety rating section of the

Monroney label: (1) crash protection information—which would be comprised of a rating

(possibly one which maintains the Agency’s 5-star ratings brand) that is tied to a vehicle’s

performance in NCAP crashworthiness and rollover testing; (2) safety technology information—

which could be comprised of a rating (possibly one that uses the Agency’s 5-star ratings brand, a

three-tier medal award system, or points) that is tied to a vehicle’s ability to avoid a crash based

210

49 CFR part 575, Section 302, “Vehicle labeling of safety rating information (compliance required for model

year 2012 and later vehicles manufactured on or after January 31, 2012),” specifies that the safety ratings

information landscape should be at least 4.5 in. wide and 3.5 in. tall or cover at least 8 percent of the total area of the

Monroney label—whichever is larger. Currently, any change that requires modification of the safety rating

information presented on the Monroney label would require a notice and comment rulemaking action pursuant to the

Administrative Procedure Act.

161

on its performance in ADAS testing conducted by NCAP; and (3) overall vehicle safety

performance information—which could give recognition to vehicles that are top performers in

both the crash protection and safety technology information categories for a given model year.

NHTSA believes that efforts to develop a label that incorporates these three main

overarching categories—crash protection information, safety technology information, and overall

vehicle safety performance information—should also strive to reduce the need to update the

Monroney label by way of rulemaking when future changes are made to the NCAP program.

NHTSA intends to develop potential label changes by conducting consumer research. In

the past, NCAP has benefitted from research on the illustration of NCAP vehicle safety

information in the safety rating section of the Monroney label. NHTSA plans to conduct

qualitative and quantitative consumer market research to: (1) evaluate the overall appeal of the

safety rating label concept mentioned above and identify specific likes and dislikes associated

with each of the three main categories on the label; (2) measure the ease of comprehension for

the safety rating label concept and understand which visual and text features are most effective at

conveying vehicle safety information; (3) assess the distinctiveness of how the information is

displayed and understand how best to make the vehicle safety information stand out on the

Monroney label; and (4) identify additional areas of improvement related to the three potential

main label categories relating to crash protection information, safety technology information, and

overall vehicle performance information.

211

NHTSA plans to use the results of this research to

determine how best to convey safety rating information to the public.

VI. Establishing a Roadmap for NCAP

The fourth part to this notice discusses, for the first time in NCAP, a roadmap that sets

forth NHTSA’s plans for upgrading NCAP over the next several years. As mentioned at the

211

NHTSA published a notice on April 28, 2020, seeking public comment on the information collection aspect of

the consumer market research.

162

beginning of this notice, the Agency’s efforts outlined herein include both NHTSA’s near- and

long-term strategies for upgrading NCAP.

Fulfillment of the roadmap will involve NHTSA’s issuing planned proposed upgrades in

phases as vehicle safety-related systems and technologies mature and data about their use and

efficacy become known. The systems and technologies would include new vehicle-based

crashworthiness and crash avoidance systems as well as systems-based improvements, such as

occupant restraints and headlamp system performance upgrades. NHTSA would issue a final

decision document following an RFC that responds to comments and provides appropriate lead

time. This phased process allows stakeholders to provide data and views on proposed program

updates, and allows NHTSA more flexibility to pursue program updates quicker.

Since 2015, NHTSA has worked to finalize its research on pedestrian crash protection

(head, and upper and lower leg impact tests), advanced anthropomorphic test devices (crash test

dummies) in frontal and side impact tests, a new frontal oblique crash test, and an updated

rollover risk curve. NHTSA has included these initiatives in the mid-term component of the 10-

year roadmap because the Agency reasonably believes they would meet the four prerequisites for

inclusion in NCAP.

212

Initiatives in the mid-term component of the 10-year roadmap identify

and prioritize safety opportunities and technologies that are practical and for which objective

tests and criteria, and other consumer data exist.

213

In addition to the items in the roadmap discussed below, NHTSA is taking an

unprecedented step to consider expanding NCAP to include safety technologies that may have

the potential to help drivers make safe driving choices, as discussed in the next section. This

aspect of NCAP would focus on the relationship between technology and behavioral safety, and

212

The four requisites are: (1) the technology addresses a safety need; (2) system designs exist that can mitigate the

safety problem; (3) the technology provides the potential for safety benefits; and (4) a performance-based objective

test procedure exists that can assess system performance.

213

Pub. L 117-58, Sec. 24213.

163

would provide comparative information on devices that can shift driver behavior that contribute

to crashes (e.g., speeding, and drowsy-, impaired- and distracted-driving). Initiatives on these

technologies could be woven into both the first and second half (i.e., long-term portion) of the

10-year roadmap, depending on whether the technologies and objective tests and criteria are

sufficiently developed to meet NHTSA’s four prerequisites for inclusion in NCAP. Initiatives in

the long-term component of the roadmap include an identification of any safety opportunity or

technology not included in the mid-term component for a variety of reasons, and those initiatives

that would most benefit from stakeholder input and comments from the public. The Agency

believes the plans outlined below would fulfill the requirements set forth in Section 24213 of the

Bipartisan Infrastructure Law for the 10-year New Car Assessment Program roadmap once this

RFC is finalized.

The Bipartisan Infrastructure Law requires that NHTSA establish a roadmap for the

implementation of NCAP not later than one year after the law’s enactment.

214

This roadmap

must cover a term of ten years, consisting of a mid-term component and a long-term

component.

215

This roadmap aligns with relevant Agency priorities, performance plans, agendas,

and any other relevant NHTSA plans.

216

Additionally, the contents of the roadmap must include a plan for any changes for NCAP,

which includes descriptions of actions to be carried out and shall, as applicable, incorporate

objective criteria for evaluating safety technologies and reasonable time periods for changes to

NCAP that include new or updated tests.

217

NHTSA has long-established criteria for evaluating

safety technologies for inclusion in NCAP, which is discussed in detail earlier in this notice and

in several previous notices. NHTSA also uses the notice and comment period to ensure the time

214

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(b).

215

Id.

216

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(c)(2)(A).

217

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(c)(1)(A).

164

periods for changes to NCAP are reasonable, and the Agency expects this practice to continue.

As part of the Agency’s development of next steps for NCAP, NHTSA regularly evaluates other

rating systems within the United States and abroad, including whether there are safety benefits of

consistency with those other rating systems.

218

There are other benefits for being consistent, but

safety is NHTSA’s, and thus, NCAP’s, top priority.

Next, the roadmap shall include key milestones, including the anticipated start of an

action, completion of an action, and effective date of an update.

219

While NHTSA can

reasonably anticipate when the start of actions may occur in the mid-term portion of the

roadmap, many technologies in the long-term portion of the roadmap will require additional

research, test procedure development, product development and maturity, and a number of other

factors that prevent the Agency from providing more detail on the anticipated start of an action.

As such, NHTSA can only provide the estimated start date of 2025-2031. Completion of action

is highly dependent upon the notice and comment process, and the effective date would be highly

dependent on the completion of an action. Completion dates are dependent on the number and

depth of the comments received in response to an RFC, along with the technical research

necessary to resolve any challenging issues in the comments. Effective dates are dependent on

completion dates. As such, NHTSA cannot reasonably anticipate those timelines in advance.

The Bipartisan Infrastructure Law also requires that the mid-term portion of the roadmap

identify and prioritize safety opportunities and technologies that are practical and for which

objective rating tests, evaluation criteria, and other consumer data exist.

220

In the mid-term

portion of the roadmap, NHTSA has included only those technologies that are practical and that

otherwise meet the requirements in the law. With respect to the long-term portion of the

roadmap, NHTSA must identify and prioritize safety opportunities and technologies that exist or

218

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(c)(4).

219

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(c)(1)(B).

220

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(c)(2)(A).

165

are in development.

221

NHTSA has met both of these requirements in the following sections,

prioritizing safety opportunities and technologies that are practical and for which objective rating

tests, evaluation criteria, and other consumer data exist in the mid-term portion, and identifying

safety opportunities and technologies that exist or are in development in the long-term portion.

Any safety opportunity or technology not included in this roadmap was omitted because

NHTSA is not considering inclusion in NCAP at this time.

222

In the next five years, addition of

other technologies or opportunities to the roadmap would be subject to NHTSA’s four

prerequisites for inclusion in NCAP, the requirements of the Bipartisan Infrastructure Law for

inclusion in any part of the roadmap, and the appropriateness of the technology or opportunity

for a consumer information program.

Per Sec. 24213(c), NHTSA must request comment on the roadmap and review and

incorporate these comments, as appropriate.

223

This RFC requests comments from the public on

the roadmap. NHTSA considers the notice and comment process to be the primary form of

stakeholder engagement, though the Agency reserves the right to conduct other forms of

engagement to ensure that input received represents a diversity of technical background and

viewpoints.

224

With regard to a roadmap, NHTSA requests feedback on the following:

(59) identification of safety opportunities or technologies in development that could be

included in future roadmaps,

(60) opportunities to benefit from collaboration or harmonization with other rating

programs, and

(61) other issues to assist with long-term planning.

2021-2022 Timeframe

221

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(c)(2)(B).

222

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(c)(3).

223

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(e).

224

Pub. L 117-58, Sec. 24213(c)(1); 49 USC § 32310(d).

166

 As discussed in detail in this notice, NHTSA proposes to add four new ADAS

technologies (LKS, BSD, BSI, and PAEB) in NCAP.

 In addition to improving the safety and protection of motor vehicle occupants,

NHTSA continues its efforts and focus to improve the safety of pedestrians and

vulnerable road users. NHTSA plans to propose a crashworthiness pedestrian

protection testing program in NCAP in 2022. The pedestrian protection program

would incorporate three crashworthiness tests (i.e., head-to-hood, upper leg-to-

hood leading edge, and lower leg-to-bumper) discussed in the December 2015

RFC.

225

A crashworthiness pedestrian protection testing program would measure

how well passenger cars, trucks, and sport utility vehicles protect pedestrians in

the event of a crash. The program would further complement the safety achieved

by pedestrian automatic emergency braking by measuring the safety performance

of new vehicles to pedestrian impacts and encouraging safer vehicle designs for

pedestrians.

2022-2023 Timeframe

 NHTSA plans to propose using the THOR-50M in NCAP’s full frontal impact

tests and the WorldSID-50M in the program’s side impact barrier and side impact

pole tests soon after work commences to add the dummies to 49 CFR part 572

and FMVSSs.

226

The Agency would inform the public (in request for comment

notices) how these crash test dummies would be utilized in various NCAP test

modes.

225

80 FR 78521 (Dec. 16, 2015), pp. 78547-78550.

226

NHTSA included new rulemakings in the Spring 2020 Regulatory Agenda that would adopt the THOR-50M and

WorldSID-50M into NHTSA’s regulation for anthropomorphic test devices, 49 CFR part 572

(https://www.reginfo.gov, RIN 2127-AM20 and https://www.reginfo.gov, RIN 2127-AM22, respectively). NHTSA

also included rulemakings that would adopt use of the THOR-50M and WorldSID-50M at the manufacturers’ option

in NHTSA compliance tests for FMVSS No. 208, “Occupant crash protection,” (https://www.reginfo.gov, RIN

2127-AM21) and FMVSS No. 214, “Side impact protection,” (https://www.reginfo.gov, RIN 2127-AM23),

respectively.

167

 In the December 2015 notice, NHTSA announced it would like to include a

frontal oblique crash test in NCAP.

227

In response to that notice, commenters

requested that the Agency provide the public with additional information on the

target population as well as costs and benefits. They also argued that

countermeasure studies have not been completed and questioned the repeatability

and reproducibility of both the test procedure and the oblique moving deformable

barrier. NHTSA has continued its frontal oblique research and kept the public

informed of its findings.

228

A cornerstone of the procedure is the use of THOR-

50M dummies in the driver and right front passenger positions. NHTSA plans to

determine in 2022 whether this new crash test mode is appropriate for inclusion in

an FMVSS and/or NCAP. If a determination is made to include the test in NCAP,

the notice and comment process would follow soon thereafter.

 NHTSA will consider incorporating several additional advanced crash avoidance

technologies including lighting systems for improved nighttime pedestrian

visibility into NCAP in the near future, and will be announcing next steps during

this timeframe. These include: (1) adaptive driving beam headlights; (2)

upgraded lower beam headlighting; (3) semiautomatic headlamp beam-switching;

and (4) rear automatic braking for pedestrian protection.

2023-2024 Timeframe

227

80 FR 78521 (Dec. 16, 2015), pages 78530 through 78531;

https://one.nhtsa.gov/Research/Crashworthiness/Small%20Overlap%20and%20Oblique%20Testing

228

See www.regulations.gov, Docket No. NHTSA-2020-0016 for document Repeatability and Reproducibility of

Oblique Moving Deformable Barrier Test Procedure (Saunders 2018); Saunders, J. and Parent, D., “Repeatability

and Reproducibility of Oblique Moving Deformable Barrier Test Procedure,” SAE

Technical Paper 2018-01-1055, 2018, doi:10.4271/2018-01-1055; https://rosap.ntl.bts.gov/view/dot/41934

Structural Countermeasure Research Program; https://www.nhtsa.gov/crash-simulation-vehicle-models Vehicle

Interior and Restraint Modeling and Structural Countermeasure Research Program sections.

168

 A multi-year consumer research effort is underway to modernize the vehicle

safety rating section of the Monroney label. Once the consumer research is

complete, the Agency plans to begin a rulemaking action in 2023 to update the

Monroney label with a new labeling concept.

 Also in 2023, NHTSA plans to commence revising its 5-star safety ratings

system. The Agency has sought comment on several approaches to provide

consumers with vehicle safety ratings that provide more meaningful safety

information and discriminate performance of vehicles among the fleet. NHTSA

discusses this issue in detail in a section below.

2025-2031 Timeframe

In NHTSA’s long-term component of the roadmap, NHTSA includes a variety of

technologies and foci that attempt to overcome many safety challenges for which the

technologies available may not be as mature or may warrant additional study from NHTSA.

NHTSA is seeking stakeholder input on the appropriateness of each of these technologies for the

program and whether commenters believe that these technologies will meet the program’s four

prerequisites within the next 5- or 10-year time frame.

NHTSA will be further assessing and developing tests for the following crash avoidance

technologies: (1) intersection safety assist; (2) opposing traffic safety assist; and (3) automatic

emergency braking for all vulnerable road users (including bicyclists and motorcyclists) in all

major crash scenarios including when the vehicle is turning left or right. NHTSA will also be

assessing the effectiveness of systems that are or will become available in the fleet. The Agency

hopes that information will be available that would support a proposal in 2025 or beyond to

include these three technologies in NCAP.

169

Based on comments received from stakeholders, if a technology development is mature

and the available data in the next several years meet the Agency’s four prerequisites, NHTSA

would issue a proposal for inclusion in NCAP during the five-year mid-term timeline.

VII. Adding Emerging Vehicle Technologies for Safe Driving Choices

NCAP has traditionally focused on crashworthiness technologies that protect the vehicle

occupants in the event of a collision. The more advanced ADAS technologies that are the focus

of this notice take the next step and provide technologies that can assist drivers, or in certain

cases correct drivers’ action in ways that can avoid or mitigate crashes. NHTSA has also begun

to consider ways NCAP could be used to encourage technologies that protect road users other

than the vehicles occupants, such as pedestrians and pedalcyclists.

As beneficial as these technologies may be, NHTSA recognizes that risky driving

behaviors and poor driver choices continue to amplify crash, injury, and fatality risks on our

roadways. Accordingly, NHTSA is interested in safety technologies that have the ability to

address the prevalent driver behaviors that contribute to roadway fatalities. For example, there

are several available and emerging safety technologies that have the potential to address

speeding and drowsy-, impaired-, distracted-, and unbelted-driving, thereby reducing the risk of

crashes that lead to injury or death, which are the subjects of analysis, research, and examination.

NHTSA is exploring opportunities to encourage the development and deployment of

these technologies. While more must be known about the effectiveness and consumer

acceptance of these systems, NHTSA strongly believes that these technologies will mature and

show efficacy. In the nearer term, then, the Agency sees potential in highlighting vehicles

equipped with these technologies on its website, and possibly elsewhere, to improve public

awareness, and encourage vehicle manufacturer development and adoption. NHTSA will

conduct research to develop objective test procedures and criteria to evaluate the performance

and effectiveness of these technologies. Initiatives on these technologies would be woven into

170

both the first and second half (i.e., long-term portion) of the 10-year roadmap, depending on

whether the technologies and objective tests and criteria are sufficiently developed to meet

NHTSA’s four prerequisites for inclusion in NCAP.

A. Driver Monitoring Systems

Driver monitoring systems use a variety of sensors and software to detect and/or infer

driver state based on estimation approaches. For example, certain types of driver monitoring

systems have shown promise in detecting the state of a driver’s drowsiness.

229

As vehicle

technologies have evolved, driver monitoring systems have been more commonly introduced and

applied to various driver states, particularly as one of the countermeasures against potential

misuse of ADAS. Currently, there are varied approaches to driver monitoring across vehicle and

equipment manufacturers.

NHTSA is considering adding driver monitoring systems as an NCAP technology to

encourage further deployment of effective driver monitoring systems into vehicles. NHTSA

seeks comment on the following to help the Agency determine whether to implement driver

monitoring systems in NCAP:

(62) What are the capabilities of the various available approaches to driver monitoring

systems (e.g., steering wheel sensors, eye tracking cameras, etc.) to detect or infer

different driver state measurement or estimations (e.g., visual attention, drowsiness,

medical incapacity, etc.)? What is the associated confidence or reliability in detecting

or inferring such driver states and what supporting data exist?

(63) Of further interest are the types of system actions taken based on a driver monitoring

system’s estimate of a driver’s state. What are the types and modes of associated

warnings, interventions, and other mitigation strategies that are most effective for

229

Brown, T., Lee, J., Schwarz, C., Fiorentino, D., McDonald, A., Traube, E., Nadler, E. (2013). Detection of Driver

Impairment from Drowsiness. 23

Enhanced Safety of Vehicles Conference, Seoul, Republic of Korea. May 2013.

Paper Number 13-0346.

171

different driver states or impairments (e.g., drowsy, medical, distraction)? What

research data exist that substantiate effectiveness of these interventions?

(64) Are there relevant thresholds and strategies for performance (e.g., alert versus some

degree of intervention) that would warrant some type of NCAP credit?

(65) Since different driver states (e.g., visual distraction and intoxication) can result in

similar driving behaviors (e.g., wide within-lane position variability), comments

regarding opportunities and tradeoffs in mitigation strategies when the originating

cause is not conclusive are of specific interest.

(66) What types of consumer acceptance information (e.g., consumer interest or feedback

data) are available or are foreseen for implementation of different types of driver

monitoring systems and associated mitigation strategies for driver impairment,

drowsiness, or visual inattention? Are there privacy concerns? What are the related

privacy protection strategies? Are there use or preference data on a selectable feature

that could be optionally enabled by consumers (e.g., for teen drivers by their parents)?

B. Driver Distraction

According to NHTSA’s statistics, driver distraction resulted in at least 3,000 known

deaths in 2019.

230

Often discussions regarding distracted driving center around cell phone use

and texting, but distracted driving also includes other activities such adjusting the radio or

climate controls or accessing other in-vehicle systems. In-vehicle devices and Human-Machine

Interfaces (HMI) can be strategically designed to avoid or limit opportunities for driver

distraction.

231

Easy access to manual controls in traditional or expected locations can minimize

230

National Center for Statistics and Analysis. (2020, December). Overview of Motor Vehicle Crashes in 2019.

(Traffic Safety Facts. Report No. DOT HS 813 060). Washington, DC: National Highway Traffic Safety

Administration.

231

In 2013, NHTSA published “Visual-Manual NHTSA Driver Distraction Guidelines for In-Vehicle Electronic

Devices.” These voluntary guidelines apply to original equipment in-vehicle electronic devices used by the driver to

perform secondary tasks (communications, entertainment, information gathering, navigation tasks, etc. are

considered secondary tasks) through visual-manual means.

172

the amount of time a driver’s eyes are off the road and hands are off the steering wheel, as well

as the time needed for the driver to activate the control quickly in time-critical traffic conflict

scenarios (e.g., a driver reaches to activate the horn button in a crash-imminent situation, but

finds that the control of horn activation is not in the expected, typical location).

NHTSA seeks comment on the following:

(67) What in-vehicle and HMI design characteristics would be most helpful to include in an

NCAP rating that focuses on ease of use? What research data exist to support

objectively characterizing ease of use for vehicle controls and displays?

(68) What are specific countermeasures or approaches to mitigate driver distraction, and

what are the associated effectiveness metrics that may be feasible and appropriate for

inclusion in the NCAP program? Methods may include driver monitoring and action

strategies, HMI design considerations, expanded in-motion secondary task lockouts,

phone application/notification limitations while paired with the vehicle, etc.

(69) What distraction mitigation measures could be considered for NCAP credit?

C. Alcohol Detection

Alcohol-impaired driving continues to be a pervasive contributing factor to roadway

fatalities, with over 10,000 deaths in the U.S. in 2019.

232

NHTSA has explored many ways in

which alcohol-impaired driving risks can be effectively mitigated both through vehicle

technologies and strategic public outreach and enforcement.

233

In 2020, NHTSA published a

Request for Information notice seeking input on Impaired Driving Technologies in the Federal

234

Specifically, the notice requested information on available or late stage technology

under development for impaired driving detection and mitigation. A total of 12 comments were

https://www.federalregister.gov/documents/2013/04/26/2013-09883/visual-manual-nhtsa-driver-distraction-

guidelines-for-in-vehicle-electronic-devices.

232

Ibid.

233

NHTSA has researched the Driver Alcohol Detection System for Safety (DADSS) program.

234

85 FR 71987 (November 12, 2020).

173

received.

235

Comments were submitted about emerging technologies that can directly measure

impairment though blood alcohol concentration at the beginning of a trip as well as technologies

that infer alcohol impairment through a combination of driver monitoring and other vehicle

sensors tracking during the course of a trip.

NHTSA seeks comment on the following aspects of alcohol detection systems:

(70) Are there opportunities for including alcohol-impairment technology in NCAP? What

types of metrics, thresholds, and tests could be considered? Could voluntary

deployment or adoption be positively influenced through NCAP credit?

(71) How can NCAP procedures be described in objective terms that could be inclusive of

various approaches, such as detection systems and inference systems? Are there

particular challenges with any approach that may need special considerations? What

supporting research data exist that document relevant performance factors such as

sensing accuracy and detection algorithm efficacy? 

(72) When a system detects alcohol-impairment during the course of a trip, what actions

could the system take in a safe manner? What are the safety considerations related to

various options that manufacturers may be considering (e.g., speed reduction,

performing a safe stop, pulling over, or flasher activation)? How should various

actions be considered for NCAP credit?

(73) What is known related to consumer acceptance of alcohol-impaired driving detection

and mitigation functions, and how may that differ with respect to direct measurement

approaches versus estimation techniques using a driver monitoring system? What

consumer interest or feedback data exist relating to this topic? Are there privacy

concerns or privacy protection strategies with various approaches? What are the related

privacy protection strategies?

235

https://www.regulations.gov/document/NHTSA-2020-0102-0001/comment.

174

D. Seat Belt Interlocks

Seat belt use in passenger vehicles saved an estimated 14,955 lives in 2017.

236

The

national seat belt use rate in the United States was 90.7 percent in 2019.

237

Among the 22,215

passenger vehicle occupants killed in 2019, almost half (47 percent) were unrestrained. For

those passenger vehicle occupants who survived crashes where someone else died, only 14

percent were unrestrained compared to 47 percent of those who died.

238, 239

Currently, NHTSA uses an array of countermeasures, including the Click It or Ticket

campaign and State primary enforcement laws, to encourage seat belt use. The Agency requires

seat belt reminders for the driver’s seat.

240

As of the 2018 model year, about 95 percent of

vehicles voluntarily offer front passenger warnings. NHTSA also informs consumers searching

for vehicle ratings on www.NHTSA.gov as to the availability of optional front passenger and

rear seat belt reminder systems, which typically provide a visual and auditory warning to the

driver at the onset of a trip and if a passenger unbuckles during a trip.

Methods for detecting seat belt misuse have advanced in recent years. A 2018 NHTSA

report, “Performance Assessment of Prototype Seat Belt Misuse Detection System,” showed that

the system correctly identified seat belt misuse in 95 percent of trials on average across multiple

common seat belt misuse scenarios.

241

This type of seat belt misuse or non-use detection could

be coupled with various types of seat belt interlock systems to encourage seat belt use. Although

NHTSA is not aware of any such system being currently in production, various prototype

systems have been developed by manufacturers.

242

These systems could include transmission

interlock, ignition interlock, and entertainment system interlock. Such systems could prevent

236

DOT HS 812 683. Latest agency estimate available.

237

DOT HS 812 875.

238

DOT HS 813 060.

239

Based on known restraint use. Restraint use was unknown for 8.7 percent of passenger vehicle occupant

fatalities in 2019.

240

49 CFR 571.208.

241

DOT HS 812 496.

242

“NHTSA’ Research on Seat Belt Interlocks,” SAE Government Industry Meeting, January 24-26, 2018.

175

drivers from shifting into gear, starting their vehicle, or using their vehicle’s entertainment

system, respectively, if the driver and/or front passenger is unbelted. Another potential strategy

could be speed limiter interlock systems. Such a system could first issue a seat belt reminder

warning if the driver begins driving and is unbelted, and then automatically reduce vehicle speed

to a very low speed after a certain warning period if the driver remains unbelted.

NHTSA requests comment on the following related to seat belt interlock systems:

(74) Should NCAP consider credit for a seat belt reminder system with a continuous or

intermittent audible signal that does not cease until the seat belt is properly buckled

(i.e., after the 60 second FMVSS No. 208 minimum)? What data are available to

support associated effectiveness? Are certain audible signal characteristics more

effective than others?

(75) Is there an opportunity for including a seat belt interlock assessment in NCAP?

(76) If the Agency were to encourage seat belt interlock adoption through NCAP, should all

interlock system approaches be considered, or only certain types? If so, which ones?

What metrics could be evaluated for each? Should differing credit be applied

depending upon interlock system approach?

(77) Should seat belt interlocks be considered for all seating positions in the vehicle, or only

the front seats? Could there be an opportunity for differentiation in this respect?

(78) What information is known or anticipated with respect to consumer acceptance of seat

belt interlock systems and/or persistent seat belt reminder systems in vehicles? What

consumer interest or feedback data exist on this topic?

(79) Could there be an NCAP opportunity in a selectable feature that could be optionally

engaged such as in the context of a “teen mode” feature?

E. Intelligent Speed Assist

176

Speeding continues to be one of the critical factors in fatal crashes on American

roadways. Specifically, driving too fast for conditions and exceeding the posted limit are two

prevalent factors that contribute to traffic crashes. For more than two decades, NHTSA has

identified speed as being a factor in at least nearly one-third of all motor vehicle related fatalities.

For example, in 2019, of the 36,096 traffic-related fatalities occurred on U.S. roadways, 9,478 of

those were positively identified as speeding-related.

243

These totals may underreport speeding,

potentially to a significant degree, as they are based on whether any driver in the crash was

charged with a speeding-related offense or if a police officer indicated that racing, driving too

fast for conditions, or exceeding the posted speed limit was a contributing factor in the crash. As

this reporting is based on aggregated police actions rather than an engineering analysis of

individual crashes, it may tend to underestimate the presence of speeding, particularly in crashes

where the speeding was not clearly obvious but still a factor in either the occurrence or severity

of the crash.

Too few drivers view speeding as an immediate risk to their personal safety or the safety

of others, including pedestrians and vulnerable road users. Yet, the consequences of speeding

include: greater potential for loss of vehicle control; reduced effectiveness of occupant protection

equipment; increased stopping distance after the driver perceives a danger; increased degree of

crash severity leading to more severe injuries; economic implications of a speed-related crash;

and increased fuel consumption and cost. The probability of death, disfigurement, or debilitating

injury grows with higher speed at impact.

NHTSA engages with State and local jurisdictions as well as national law enforcement

partners to provide funding and educational materials which address speeding. Speed limiter

features, which prevent a vehicle from traveling over a certain speed by limiting engine power,

243

Traffic Safety Facts 2019 “A Compilation of Motor Vehicle Crash Data.” U.S. Department of Transportation.

National Highway Traffic Safety Administration.

177

are available in the U.S. market and widely used in heavy-duty tractor-trailers and other fleet-

based vehicles. In addition, nearly all vehicles are equipped with a mechanism that limits their

top-end speed, even if that speed is quite high. These systems either prevent a vehicle from

exceeding a preset specific speed regardless of location, or they use GPS and/or camera data to

determine the speed limit of the current road and apply mitigation measures to reduce speeding.

Vehicles equipped with an intelligent speed assist system can display the current speed limit to

the driver at all times. Should the driver exceed the speed limit for the road, the system can

provide a visual or auditory alert or actively slow the vehicle to an appropriate speed. Typically,

many existing intelligent speed assist systems can be temporarily overridden by the driver by

depressing the accelerator pedal firmly.

NHTSA is committed to addressing this important safety issue to further reduce fatalities

and injuries. NHTSA requests comment on the following aspects of intelligent speed assist

systems in passenger vehicles as well as other approaches that are not discussed in this notice.

(80) Should NHTSA take into consideration systems, such as intelligent speed assist

systems, which determine current speed limits and warn the driver or adjust the

maximum traveling speed accordingly? Should there be a differentiation between

warning and intervention type intelligent speed assist systems in this consideration?

Should systems that allow for some small amount of speeding over the limit before

intervening be treated the same or differently than systems that are specifically keyed to

a road’s speed limit? What about for systems that allow driver override versus systems

that do not?

(81) Are there specific protocols that should be considered when evaluating speed assist

system functionality?

178

(82) What information is known or anticipated with respect to consumer acceptance of

intelligent speed assist systems? What consumer interest or feedback data exist on this

topic?

(83) Are there other means that the Agency should consider to prevent excessive speeding?

F. Rear Seat Child Reminder Assist

Data indicate that since 1998, nearly 900 children (an average of 38 per year) have died

in the U.S. of hyperthermia (vehicular heatstroke) because they were left or became trapped in a

hot vehicle. 2018 and 2019 saw a record number of vehicular heatstroke related deaths at 53

each year.

244

Children were in the vehicles due to a variety of circumstances – some gain entry

to a parked vehicle, whereas over 50 percent are forgotten in the vehicle by caregivers.

245

To address these tragedies, many companies have developed aftermarket devices to

remind parents and caregivers that a child may be left inside the vehicle. NHTSA has assessed

several products and developed a test methodology for evaluating future products.

246

NHTSA

subsequently opened a public docket inviting all interested parties to submit information

regarding efforts or technological innovations to help prevent vehicular heatstroke.

247

Also,

NHTSA has media campaigns, such as “Where’s Baby? Look Before You Lock,” to raise

awareness to parents and caregivers on the dangers of vehicular heatstroke.

In recent years, in-vehicle rear seat child reminder technology has been introduced into a

number of vehicle makes and models. Many of these technological solutions utilize “door logic”

to determine if there is potentially a child in the rear seat of the vehicle. The vehicle door logic

checks to see if the rear seat doors were opened and closed at the start of the trip and then

displays a reminder in the dash board with an audio cue for the driver to check the back seat

244

www.noheatstroke.org

245

Id.

246

Rudd, R., Prasad, A., Weston, D., & Wietholter, K. (2015, July). Functional assessment of unattended child

reminder systems. (Report No. DOT HS 812 187). Washington, DC: National Highway Traffic Safety

Administration.

247

https://www.regulations.gov/docket?D=NHTSA-2019-0126

179

when the vehicle is turned off. In September 2019, the Alliance of Automobile Manufacturers

and the Association of Global Automakers (now collectively known as the Alliance for

Automotive Innovation) announced that a voluntary agreement had been formed by its member

companies to incorporate rear seat child reminder systems into their vehicles as standard

equipment no later than the 2025 model year.

248

NHTSA requests comment on the following issues related to rear seat child reminder

systems designed to prevent vehicular heatstroke.

(84) If NHTSA considers this technology for inclusion in NCAP, are door logic solutions

sufficient? Should NHTSA only consider systems that detect the presence of a child?

(85) What research data exist to substantiate differences in effectiveness of these system

types?

(86) Are there specific protocols that should be considered when evaluating these in-vehicle

rear seat child reminder systems?

(87) What information is known or anticipated with respect to consumer acceptance of

integrated rear seat child reminder systems in vehicles? What consumer interest or

feedback data exist on this topic? 

VIII. Revising the 5-Star Safety Rating System

NHTSA is seeking comment on several approaches to provide consumers with vehicle

safety ratings that provide more meaningful safety information and provide consumers with more

ways to determine relative performance of vehicles among the fleet. In the current 5-star safety

ratings system, as described in detail in the July 2008 final decision notice, injury readings

recorded from crash test dummies used in NCAP’s frontal impact, side impact barrier, and side

impact pole tests are assessed using injury risk curves designed to predict the chance of a

248

https://www.autosinnovate.org/safety/heatstroke/Automakers%20Commit%20to%20Helping%20Combat%20Chi

ld%20Heatstroke.pdf

180

vehicle’s occupant receiving similar injuries.

249

For each occupant in each crash test, the risks of

injury to each body region assessed are combined to produce a combined probability of injury to

each occupant. The combined probabilities of injury for each occupant are divided by a

predetermined baseline risk of injury. This baseline risk of injury approximates the fleet average

injury risk for each crash test. Dividing each combined occupant probability of injury by the

baseline risk of injury results in a relative assessment of that occupant’s combined injury risk

versus a known fleet average. These calculations result in six summary scores for each vehicle

representing the relative risk of injury for the following occupants: (1) the driver and front seat

passenger in the frontal impact test; (2) the driver and rear seat passenger in the side impact

barrier test; (3) the driver in the side impact pole test; and (4) the relative risk for all occupants in

rollovers with respect to a baseline injury risk. These relative risks are then converted to star

ratings to help consumers make informed vehicle purchasing decisions.

NHTSA seeks public comment on a few potential concepts it could use to develop a new

5-star safety ratings system in the future. Some areas of consideration discussed below could be

used in conjunction with one another, while others could work better as standalone options.

Ideally, any future 5-star safety ratings system should not only fulfill the program mission, but

also be sufficiently flexible to allow for continuing updates to NCAP to encourage further

vehicle safety improvements.

A. Points-Based Ratings System Concept

NHTSA is seeking comment on the use of a potential points-based system to calculate

future 5-star safety ratings for the crashworthiness testing program when the Agency decides to

update that program. In this system, star ratings could be assigned directly from point values

related to the results from crash test dummies. The current system is based on a linear

combination of the probability of injury for multiple body regions, some at different severity

249

73 FR 40016 (July 11, 2008), http://regulations.gov, Docket No. NHTSA-2006-26555-0114.

181

levels, which can result in some body regions being overlooked. A point-based system, on the

other hand, would provide more flexibility to target injury criteria more representative of real-

world injury incidence. The Agency believes that this potential method would provide more

flexibility in the future when updating the program through a phased approach. For instance,

new testing devices (e.g., crash test dummies), procedures, injury measurements, or other criteria

could be added to the 5-star-ratings system. Points could be based on critical injury risk curve

values or on criteria, such as reference values from existing Federal regulations or other Agency

data.

This points-based rating system approach would be similar to those used in other vehicle

safety consumer information programs such as IIHS and Euro NCAP. Upper and lower

performance targets would be established for each test dummy body region assessed in crash

tests. Maximum points would be awarded if Injury Assessment Reference Values (IARVs) meet

the lower target or better. A linearized number of points would be awarded for injury assessment

values that are between the lower and upper targets. No points would be assigned for those that

exceed the upper target for the respective body region (or perhaps the entire occupant). Risk

curves would no longer be used exclusively to calculate a combined injury probability from the

various body regions and ultimately star ratings. Critical risk curve values, IARVs, or other

accepted injury limits would be used to establish performance targets and related points

assignments.

In addition to the injury criteria currently included in the 5-star safety ratings system, data

to support several other injury criteria are collected for Agency monitoring and consumer

information on the respective NCAP dummies (Hybrid III and ES-2re 50th percentile males,

Hybrid III and SID-IIs 5th percentile females). NHTSA is seeking comment on whether any

additional measurements that are not part of the existing 5-star ratings system are appropriate for

use in a points-based calculation of the future star ratings.

182

Currently, if measurements of certain injury criteria that are included in related FMVSSs

exceed standard limits, the Agency would assign a “safety concern” designation on its website

and on the vehicle window sticker (Monroney label).

250

If measurements of certain injury

criteria that are not part of FMVSSs exceed established limits, the Agency highlights those on its

website (but not on the Monroney label) with footnotes. In both of these cases, the Agency seeks

to inform consumers of potentially higher injury risks in body regions that are not captured by

the existing 5-star safety ratings system. The Agency recognizes that consumer confusion may

result from the presentation of a vehicle with high (4- or 5-star) ratings that is also assigned a

safety concern or injury-related footnote. One potential solution to reduce confusion would be to

implement a points-based system that allows the Agency to include the assessment of all injuries

within the calculation of the star rating, even those that may not have associated risk curves.

Thus, the Agency is seeking comment on the appropriate method.

Furthermore, NHTSA is exploring several options regarding the distribution of points

across a potential points-based ratings system. Real-world data could be used to apportion the

total number of available points to each crash mode, dummy, and/or injury value according to

severity or prevalence in the field. Alternatively, each dummy or injury value could be allotted

the same number of points, effectively normalizing each dummy or injury.

B. Baseline Risk Concept

Support for adjusting the baseline risk value associated with 5-star safety ratings has been

mixed in the past, with some in favor and others advising against it.

251

As mentioned earlier, the

Agency is again seeking comment on whether the baseline risk concept should be preserved

when considering updates to its 5-star safety ratings system in the future.

250

Id.

251

This is based on comments by participants in the October 1, 2018 public meeting and respondents to the related

docket https://www.regulations.gov/docket?D=NHTSA-2018-0055.

183

With the July 2008 final decision establishing the existing 5-star safety ratings system,

the concept of a relative star rating system was introduced for the first time.

252

As discussed

previously, after injury readings from various body regions are converted to combined

probabilities of injury risks, those combined probabilities are divided by a baseline (or average)

risk of injury that is an approximation of the vehicle fleet average injury risk. Star ratings

generated in NCAP today are a measure of how much more (or less) occupant protection the

vehicle affords when compared to an “average” vehicle.

The intent of the baseline risk as described in the July 2008 notice was to update its value

at regular intervals so that, as the average risk of injury decreased over time, ratings could

become more stringent without changing the underlying criteria. In practice, the baseline risk

has never been adjusted, which results in recent star ratings being assigned using an older

benchmark less representative of current vehicle safety levels.

253

C. Half-Star Ratings

In the December 2015 notice, the Agency sought comments on the merits of providing

ratings to consumers in half-star increments. Commenters were generally supportive of the

notion. In this notice, NHTSA continues to seek comment on whether the Agency should

disseminate its 5-star safety ratings with half-star increments. This approach could allow better

discrimination of vehicle performance for consumer information purposes by creating additional

levels within the existing 1-, 2-, 3-, 4-, and 5-star levels. Though the Agency has not conducted

consumer research on this potential approach, NHTSA believes that the public is familiar with

the general impression of half-star ratings as it is commonly found in other consumer product

rating schemes.

252

Prior to the 2010 program enhancements, NCAP star ratings were based on an absolute, independent scale of

combined injury probability. That is, the combined probability of injury from a given occupant was converted

directly into a star rating with no intermediate calculation except rounding.

253

Park, B., Rockwell, T., Collins, L., Smith, C., Aram, M. (2015), The enhanced U.S. NCAP: Five years later. 24th

Enhanced Safety of Vehicles Conference, Gothenburg, Sweden, June 2015, Paper Number 15–0314.

184

Future crashworthiness 5-star safety ratings systems most likely would contain more

elements on which vehicles are assessed. Thus, NHTSA believes that using half-star increments

may be necessary in future rating systems because they allow better discrimination of vehicle

safety performance. The half-star increments, depending on future Agency decisions, could

create anywhere from 9 to 11 levels

254

of discrimination for use in rating vehicles.

NHTSA could design any half-star rating system to require a vehicle to reach the

minimum threshold for receiving that rating level. Ratings in a system such as this would be

“rounded down” to the nearest half- or whole-star rating and would not be “rounded up” to the

next half- or whole-star rating.

D. Decimal Ratings

NHTSA is also seeking comments on whether it should consider assigning star ratings

using a decimal format in addition to or in place of assigning whole- or half-star ratings. The

decimal rating could be based on a conversion of NCAP test results by using a linear function

approach. For instance, in the current 5-star safety ratings system, this could be achieved by

relating a linear function to the VSS calculation and its associated ranges. In a potential future 5-

star safety ratings system, like one where the previously discussed points-based concept is used,

a decimal value could also be easily integrated. Providing NCAP ratings in decimal format

could provide consumers with an additional, high delineation method of discriminating vehicle

performance among the fleet for purchasing reasons.

Considering these ongoing Agency initiatives currently being pursued for future NCAP

upgrades, NHTSA requests comment on the following:

254

Depending on possible rating scales from 0-5 stars, 0.5-5 stars, or 1-5 stars, the amount of total distinct ratings

available would vary.

185

(88) What approaches are most effective to provide consumers with vehicle safety ratings

that provide meaningful information and discriminate performance of vehicles among

the fleet?

Specifically with regard to a points-based rating system, the Agency seeks comment on

the following:

(89) Is the use of additional injury criteria/body regions that are not part of the existing 5-

star ratings system appropriate for use in a points-based calculation of future star

ratings? Some injury criteria do not have associated risk curves. Are these regions

appropriate to include, and if so, what is the appropriate method by which to include

them?

Regarding the baseline risk concept and the general concept of relative ratings, NHTSA

is seeking comment on the following:

(90) Should a crashworthiness 5-star safety ratings system continue to measure a vehicle’s

performance based on a known or expected fleet average performer, or should it return

to an absolute system of rating vehicles?

(91) Considering the basic structure of the current ratings system (combined injury risk), the

potential overlapping target populations for crashworthiness and ADAS program

elements, as well as other potential concepts mentioned in this document such as a

points-based system, what would the best method of calculating the vehicle fleet

average performance be?

(92) Should the vehicle fleet average performance be updated at regular intervals, and if so,

how often?

(93) What is the most appropriate way to disseminate these updates or changes to the

public?

186

Considering a change in approach to how to present star ratings to the public, NHTSA

seeks comment on the following:

(94) Should the Agency disseminate its 5-star ratings with half-star increments?

(95) Should the Agency assign star ratings using a decimal format in addition to or in place

of whole- or half-stars?

E. Rollover Resistance Testing Program

Currently, there are two rollover resistance tests that the Agency conducts and are part of

the existing 5-star safety ratings system. The first component of this assessment is the static

measurement of the vehicle’s center of gravity height and the track width to determine the

vehicle’s static stability factor. The second component of this assessment is the dynamic

rollover test (Fishhook test) that simulates a driver taking a panic steering action in a loss-of-

control situation. The Agency uses two formulas (no tip-up and tip-up results) for calculating the

risk of rollover and then assigns a rollover rating based on the risk. NHTSA sought comment on

the approach published in the December 2015 notice to recalculate its current rollover risk curve

given the full implementation of electronic stability control (ESC) systems as standard

equipment in all vehicles manufactured on or after September 1, 2011. Commenters who

responded to the December 2015 notice were generally supportive of the Agency’s desire to

update the rollover risk curve to reflect the role of ESC deployment. However, few specific

comments on the appropriateness of the approach that was described in the notice were received

at the time.

NHTSA is not proposing changes to its two existing rollover resistance tests at this time.

However, when the Agency proposes changes to the existing 5-star ratings system, it may be

feasible to consider an update to how it assesses the rollover resistance testing component. Thus,

the Agency is seeking comment on whether any future overall vehicle ratings should continue to

include rollover resistance evaluations. Also, if the Agency updates the rollover risk curve,

187

suggestions on how to transition that data into a future overall vehicle rating would be

encouraged. The Agency expects that any future overall vehicle ratings would, at minimum,

require reweighting the contribution of each test mode to that overall rating and thus the need to

determine the most appropriate program area to include the rollover resistance tests.

(96) Should the Agency continue to include rollover resistance evaluations in its future

overall ratings?

IX. Other Activities

A. Programmatic Challenges with Self-Reported Data

Since model year 2011, vehicle manufacturers have been reporting to NHTSA their

internal test data that show whether vehicles equipped with the recommended ADAS

technologies pass NCAP’s system performance test requirements in order to receive credit from

the Agency. NHTSA assesses the information provided and then assigns check marks for

systems whose conformance with NCAP’s performance test requirements are supported by the

data. As the Agency stated in its July 2008 final decision notice, commenters were generally

supportive of NHTSA’s plan to use self-reported data from the vehicle manufacturers, in

conjunction with its own spot-check verification testing, to determine whether vehicles met

NCAP’s system performance test requirements.

255

The process by which the Agency has

accepted self-reported ADAS technology data for recommended technologies has been crucial to

the successful administration of the program.

However, this process has not been without challenges. Throughout the administration of

the ADAS assessment program in NCAP, NHTSA has identified inconsistencies in vehicle

manufacturers’ self-reported data submissions. The Agency has determined that many of these

inconsistencies stem from unfamiliarity with NCAP’s system performance test procedures,

including the use of test targets and other parameters.

255

72 FR 3473 (Jan. 25, 2007), Docket No. NHTSA-2006-26555.

188

It is critical to maintain program credibility and public trust when accepting

manufacturers’ ADAS self-reported data and disseminating it to the public. One approach to

addressing some of the aforementioned challenges is to encourage all vehicle manufacturers to

provide NHTSA with ADAS self-reported data from an independent test facility that meets

criteria demonstrating competence in NCAP testing protocols. For instance, NHTSA’s rigorous

procurement process for awarding contracts to test laboratories provides that qualified

laboratories meet specific competence requirements.

To address the challenges mentioned above, NHTSA is considering refusing to accept

self-reported data and not posting recommendations for the vehicle’s systems on its website,

when:

 Manufacturers’ self-reported ADAS test data is provided from a test facility that is not

designated as NHTSA’s contracted test laboratory, or

 The corresponding ADAS tests are not conducted in accordance with NCAP’s testing

protocols (including test devices).

NHTSA seeks comment on the following:

(97) Considering the Agency’s goal of maintaining the integrity of the program, should

NHTSA accept self-reported test data that is generated by test laboratories that are not

NHTSA’s contracted test laboratories? If no, why not? If yes, what criteria are most

relevant for evaluating whether a given laboratory can acceptably conduct ADAS

performance tests for NCAP such that the program’s credibility is upheld?

(98) As the ADAS assessment program in NCAP continues to grow in the future to include

new ADAS technologies and more complex test procedures, what other means would

best address the following program challenges: methods of data collection, maintaining

data integrity and public trust, and managing test failures, particularly during

verification testing?

189

B. Website Updates

NHTSA uses its website and the safety rating section of the Monroney label to convey to

consumers vehicle safety information provided by NCAP. Although the Monroney label is an

important tool NHTSA uses to communicate vehicle safety ratings to consumers at the point of

sale, it has limitations:

(1) The Agency must undergo a rulemaking action to change any of its content, including

minor and non-substantive changes.

256

(2) The label is limited to a certain size, only some of which is dedicated to NCAP

information, which only allows for the communication of limited safety information.

(3) By virtue of being posted on individual vehicles, the label provides limited utility as a

comparative shopping tool unless compared to labels on vehicles in the same physical location.

Thus, NHTSA uses its website to communicate a wealth of information about vehicle

safety beyond what is displayed on the Monroney label. NHTSA has structured the information

displayed on its website to align with the structure of the Monroney label. The same

crashworthiness and rollover star ratings are shown on both the label and the website. However,

crash avoidance (ADAS technologies) recommendations are not included on the Monroney label

because they were too new to be included at the time of the most recent Monroney label update,

whereas they are provided on the website.

In light of the Monroney label limitations, increasingly complex vehicle ratings and

results, and NHTSA’s desire to communicate safety information as timely as possible, NHTSA

is considering enhancing the information on its website. However, some of these enhancements

may necessitate that the information provided on the Monroney label and website deviate from

one another in structure or in content. There are limitations on the amount of information that

256

The Agency implemented the Monroney label requirement by regulation (49 CFR 575.302) pursuant to Section

10307 of the Safe, Accountable, Flexible, Efficient Transportation Equity Act; A Legacy for Users (SAFETEA-LU).

190

can be usefully conveyed on the Monroney label, so NHTSA is currently considering placing

some information on the website alone. However, while it makes sense to provide additional

information and comparative tools on the website, NHTSA is concerned that consumers could be

confused if the information in both places is not presented in the same manner. For example, the

Monroney label is currently limited to displaying whole star ratings. If, as a result of this RFC,

NHTSA decides to improve the differentiation between vehicles by displaying star ratings on its

website using new methods like a decimal equivalent value or half-stars, such a discrepancy

between the Monroney label and the website may confuse consumers.

During the October 2018 public meeting, Consumers Union suggested that NHTSA could

provide ratings on its website in a “more granular, sortable and readily comparable manner.”

Currently, the website’s functionality allows for users to input limited search terms. For

instance, a consumer may search for all vehicles in a given model year, all vehicles of a specific

make, or vehicles with a specific model name. Consumers may then filter these results by body

style, but the current body style categories are very broad and can encompass hundreds of

models. Consumers are currently limited to viewing ten vehicle models at a time in search

results, meaning that they may need to sift through many pages of results if they are simply

browsing and do not have a particular make or model in mind. NHTSA plans to address these

issues by improving the organization and versatility of the safety ratings data presented to the

public.

Once a consumer selects a vehicle for further details, they may choose to compare up to

three vehicles, but they must input the year, make, and model of the vehicles to be compared.

NHTSA intends to make changes to its www.nhtsa.gov user interface to allow for simpler

comparisons between vehicle manufacturers and types. For example, when a consumer searches

for safety rating information for a particular make and model, similar vehicles could also be

191

shown. These vehicles could be classified according to body style. The Agency expects to make

other changes to NHTSA.gov to increase the comparability of safety information.

NHTSA continues to seek comment on the following aspects of vehicle information

provided on its website:

(99) What is the potential for consumer confusion if information on the Monroney label and

on the website differs, and how can this confusion be lessened?

(100) What types of vehicles do consumers compare during their search for a new vehicle?

Do consumers often consider vehicles with different body styles (e.g., midsized sedan

versus large sport utility)?

(101) When searching for vehicle safety information, do consumers have a clear

understanding for which vehicles they are seeking information, or do they browse

through vehicle ratings to identify vehicles they may wish to purchase?

(102) When classifying vehicles by body style, what degree of classification is most

appropriate? For example, when purchasing a passenger vehicle, do consumers

consider all passenger vehicles, or are they inclined to narrow their searches to vehicles

of a subset of passenger vehicles (e.g., subcompact passenger vehicle)?

(103) Within the context of the updates considered in this notice, what is the most important

top-level safety-related information that consumers should be able to compare amongst

vehicles? Which of these pieces of information should consumers be able to use to sort

and filter search results?

C. Database Changes

NHTSA wishes to take this opportunity to inform the public about other ways the Agency

is significantly enhancing the NCAP program. We have undertaken a considerable

developmental effort to modernize the OEM submission process and our processing of data, so

that consumer information can be provided to consumers quickly and accurately. We are not

192

requesting comment in this section but are presenting this information for the benefit of the

reader.

Each year NHTSA requests vehicle manufacturers to submit new model year vehicle

information voluntarily on new passenger cars and light trucks with gross vehicle weight ratings

of 4,536 kg (10,000 pounds) or less. This information is used by NCAP primarily for consumer

information on the Agency’s website, presentation on the vehicle window stickers, and for the

selection of new model year vehicles to be tested under NCAP.

The manner in which NHTSA and vehicle manufacturers communicate information has

changed over the years—from mailed letters and faxes to spreadsheets and emails. However,

NHTSA realized a modernized process of data submission, collection, analysis, and

dissemination is necessary due to the ever-growing list of data elements needed to support an

evolving test portfolio and diverse vehicle fleet. In the last model year alone, more than 400

makes and models of passenger vehicles were sold in the United States, thus requiring vehicle

manufacturers not only to assemble detailed new vehicle data and submit them to NHTSA, but

also NHTSA to collect, sort, and analyze tremendous amounts of information.

Managing this data has become more complex, utilizing electronic spreadsheets and

email. In addition to processing spreadsheets from more than 20 organizations, maintaining

version control, checking data for accuracy, clarifying ambiguities, sending ratings letters, and

processing requests have limited the ability of the Agency’s current IT systems in storing and

analyzing data. These limitations have been exacerbated by the incorporation of ADAS

assessments into NCAP, which accepts self-reported test data from vehicle manufacturers.

Historically, these ADAS technologies have been available in a mix of vehicles within a

technology package or trim line at the make and model level, which can cause consumer

confusion as to which vehicles have the technologies. Furthermore, as NCAP is only able to

offer consumer information details at the make and model level, the additional complexity of

193

parsing trim lines and technology packages has been overly burdensome given NHTSA’s current

resources and limitations.

NHTSA is mindful that any expansion in NCAP’s ADAS assessment program will create

a long-term need to collect considerably more data elements from vehicle manufacturers. The

current data collection process of spreadsheets and emails will not suffice to fulfill this need. To

that end, NHTSA has undertaken a multi-year, multi-phase project to modernize the way in

which NCAP communicates with and receives data from relevant stakeholders. NHTSA is

currently developing a new, secure online web portal and database that will be used to send,

receive, track, store, and process program data elements and communications.

The first phase of this online portal and database development focuses on the data

submission process from the vehicle manufacturers to NHTSA. The online web portal would

allow designated representatives from each vehicle manufacturer to submit data and

correspondence by secure and trackable means. Vehicle manufacturers would be able to have

multiple representatives contribute to and approve the data submissions, and submissions could

be done in a more dedicated and focused manner than is currently feasible with conventional

spreadsheets. The data submission application would include business rules to help vehicle

manufacturers identify invalid data or typographical errors. The database portion of the project

would allow NHTSA not only to capture and store data more efficiently, but also to manage

program functions more quickly—such as faster posting of NCAP ratings to the Agency’s

website. In addition, it would allow NCAP to determine twin and carryover status in a timelier

manner. Furthermore, the database is significantly more flexible and robust than existing

spreadsheets and would allow more accurate processing of manufacturers’ self-reported data

submitted for the ADAS assessment program as well as the side air bag out-of-position testing

program. In addition, this database would allow NCAP to review vehicle fleet trends and easily

compare and track changes in individual vehicle models from one model year to the next. This

194

phase of the project has already produced a prototype, and NHTSA has received preliminary

feedback from initial beta testing.

A second phase of the project will focus on data and correspondence between NHTSA

and its test laboratories. NCAP collects vehicle-specific test setup information from the vehicle

manufacturer and separately transmits this data to its designated test laboratory. This phase of

the project would streamline the way in which the program communicates its day-to-day

operations that include the review, transmission, and archive of test data. The result of these

upgrades would allow NCAP to schedule tests, review test data, analyze test anomalies and

failures, respond to manufacturer contests, and publish safety ratings in a timelier manner.

X. Economic Analysis

The various changes in NCAP discussed in this proposal all enable a rating system that

improves consumer awareness of ADAS safety features, and encourages manufacturers to

accelerate their adoption. This accelerated adoption of ADAS would drive any economic and

societal impacts that result from these changes, and are thus the focus of this discussion of

economic analysis. Hence, the Agency has considered the potential economic effects for ADAS

technologies proposed for inclusion in NCAP and the potential benefit of introducing a rating

system for ADAS technologies.

Unlike crashworthiness safety features, where safety improvements are attributable to

improved occupant protection when a crash occurs, the impact that ADAS technologies have on

fatality and injury rates is a direct function of their effectiveness in preventing crashes or

reducing the severity of the crashes they are designed to mitigate. This effectiveness is typically

measured by using real-world statistical data, laboratory testing, or Agency expertise.

With respect to vehicle safety, the Agency believes, as discussed in detail in this notice,

the four proposed ADAS technologies have the potential to reduce vehicle crashes and injury

severities further. As cited in this notice, researchers have conducted preliminary studies to

195

estimate the effectiveness of ADAS technologies. Although these studies have been limited to

certain models or manufacturers, which may not represent the entire fleet, they do illustrate how

these systems can provide safety benefits. Thus, although the Agency does not have sufficient

data to determine the monetized safety impacts resulting from these technologies in a way

similar to that frequently done for mandated technologies – when compared to the future without

the proposed update to NCAP, NHTSA expects that these changes would likely have substantial

positive safety effects by promoting earlier and more widespread deployment of these

technologies.

NCAP also helps address the issue of asymmetric information (i.e., when one party in a

transaction is in possession of more information than the other), which can be considered a

market failure.

257

Regarding consumer information, the introduction of a potential new ADAS

rating system is anticipated to provide consumers additional vehicle safety information (e.g.,

rating based on ADAS performance and capability as well as the types of ADAS in vehicles) as

opposed to the information provided in the current program (e.g., check mark based on ADAS

performance as pass/fail) to help them make more informed purchasing decisions by better

presenting the relative safety benefits of different ADAS technologies. NHTSA believes that the

future ADAS rating would increase consumer awareness and understanding of the safety benefits

in these technologies, and, in turn, incentivize vehicle manufacturers to offer the ADAS

technologies that lead to higher ratings across a broader selection of their vehicles. Furthermore,

as these ADAS technologies mature and become more reliable and efficient, a large portion of

vehicles equipped with such systems would achieve higher ADAS ratings, and in turn consumers

would have an increasing number of safer vehicles to choose from. There is an unquantifiable

257

See

196

value to consumers in receiving accurate and comparable performance information about those

technologies among manufacturers, makes, and models.

According to NHTSA sponsored research,

258

IIHS/HLDI predicted that the number of

vehicles equipped with ADAS technologies, including BSW and Lane Keeping Warning, will

increase substantially from 2020 to 2030 and reach near full market penetration in 2050.

Although the Agency has limited data on costs of ADAS technologies to consumers, assuming

consumer demand for safety remains high, the future ADAS rating system would likely

accelerate the full adaptation of the four technologies included in this RFC—not to mention the

four existing ones. Nevertheless, the Agency does not have sufficient data, such as unit cost and

information on how soon the full adaptation will be reached with the ADAS rating, to predict the

net increase in cost to consumers, with a high degree of certainty.

XI. Public Participation

Interested parties are strongly encouraged to submit thorough and detailed comments

relating to each of the relevant areas discussed in this notice. Please see Appendix B for a

summarized list of specific questions that have been posed in this notice. Comments submitted

will help the Agency make informed decisions as it strives to advance NCAP by encouraging

continuous safety improvements for new vehicles and enhancing consumer information.

How do I prepare and submit comments?

To ensure that your comments are filed correctly in the docket, please include the docket

number of this document in your comments.

Your comments must not be more than 15 pages long (49 CFR 553.21). NHTSA

established this limit to encourage you to write your primary comments in a concise fashion.

258

See https://www.iihs.org/media/9517c308-c8d5-42e6-80fd-

a69ecd9d2128/3aaYqQ/HLDI%20Research/Bulletins/hldi_bulletin_37-11.pdf. Bulletin Vol. 34, No. 28: September

2017, “Predicted availability and fitment of safety features on registered vehicles,” Highway Loss Data Institute.

197

However, you may attach necessary additional documents to your comments. There is no limit

on the length of the attachments.

Please submit one copy (two copies if submitting by mail or hand delivery) of your

comments, including the attachments, to the docket following the instructions given above under

ADDRESSES. Please note, if you are submitting comments electronically as a PDF (Adobe)

file, NHTSA asks that the documents submitted be scanned using an Optical Character

Recognition (OCR) process, thus allowing the Agency to search and copy certain portions of

your submissions.

How do I submit confidential business information?

If you wish to submit any information under a claim of confidentiality, you should submit

three copies of your complete submission, including the information you claim to be confidential

business information, to the Office of the Chief Counsel, NHTSA, at the address given above under

FOR FURTHER INFORMATION CONTACT. In addition, you may submit a copy (two copies if

submitting by mail or hand delivery), from which you have deleted the claimed confidential business

information, to the docket by one of the methods given above under ADDRESSES. When you send

a comment containing information claimed to be confidential business information, you should

include a cover letter setting forth the information specified in NHTSA’s confidential business

information regulation (49 CFR Part 512).

Will the Agency consider late comments?

NHTSA will consider all comments received before the close of business on the comment

closing date indicated above under DATES. To the extent possible, the Agency will also consider

comments received after that date. Please note that even after the comment closing date, we will

continue to file relevant information in the docket as it becomes available. Accordingly, we

recommend that interested people periodically check the docket for new material. You may read the

comments received at the address given above under ADDRESSES. The hours of the docket are

198

indicated above in the same location. You may also see the comments on the Internet, identified by

the docket number at the heading of this notice, at www.regulations.gov.

199

XII. Appendices

Appendix A. Target Population Statistics for Crash Scenarios

259

Table A-1: Target Population Statistics, FCW/CIB/DBS

Crash Scenarios

260

Crashes Fatalities MAIS 1-5 Injuries PDOVs

2000 Rear-End, Lead Vehicle

(LV) Stopped

1,099,868 474 561,842 1,719,177

2001 Rear-End, LV Slower 174,217 527 97,402 252,341

2002 Rear-End, LV

Decelerated

374,624 155 196,731 587,031

2003 Rear-End, Other In-lane

Vehicle Higher Speed

598 3 273 829

2009 Rear-End,

Other/Unspecified

50,105 70 24,951 77,034

2300 Rear-End Possible,

Other In-lane Vehicle

Stopped

1,842 37 839 2,510

2301 Rear-End Possible,

Other In-lane Vehicle Slower

813 6 486 1,063

2302 Rear-End Possible,

Other In-lane Vehicle

Decelerated

1,475 3 860 1,900

Combined Total 1,703,541 1,275 883,386 2,641,884

Percent of Total Crashes 29.4% 3.8% 31.5% 36.3%

259

Wang, J.-S. (2019, March), Target crash population for crash avoidance technologies in passenger vehicles

(Report No. DOT HS 812 653), Washington, DC: National Highway Traffic Safety Administration.

260

The crash scenarios referenced for the FCW/CIB/DBS target population are those that comprise the subset of the

84 mutually exclusive pre-crash scenarios analyzed by VOLPE (Report No. DOT HS 812 745) that were considered

relevant for the forward collision prevention crash category (Report No. DOT HS 812 653). Each of the 84

scenarios is assigned a pre-assigned number and is followed by a brief description.

200

Table A-2: Target Population for LDW/LKA/LCA

Crash Scenarios Crashes Fatalities MAIS 1-5 Injuries PDOVs

100 1V Rollover 1st Event 4,411 63 3,155 2,104

150 2+V Rollover 1st Event 243 3 337 197

1000 1V, Roadway Departure

(RD)

966,709 9,751 359,238 679,402

1050 2+V, Roadway Departure 43,957 1,021 32,069 55,856

1100 1V Cross Centerline/Median 8,560 75 2,910 6,214

1150 2+V Cross

Centerline/Median

3,427 106 2,678 4,239

3000 ST Opposite Dir(OD),

Head-On

32,751 2,761 37,848 23,992

3009 ST OD Forward Impact,

Other

115 11 69 135

3100 ST OD, Angle Sideswipe 62,214 1,042 38,655 86,054

3200 Head-On Possible, Other

Vehicle Encroaching OD

4,008 11 2,979 5,019

Combined Total 1,126,397 14,844 479,939 863,213

Percent of Total Crashes 19.4% 44.3% 17.1% 11.9%

Table A-3: Target Population for BSD/BSI/LCM

Crash Scenarios Crashes Fatalities MAIS 1-5 Injuries PDOVs

8000 LCM in Rear End 48,749 128 26,040 71,977

8001 LCM in ST SD

Forward Impact

212 4 62 371

8002 LCM in ST SD AS 371,504 332 129,595 651,962

8003 LCM CT VT SD 58,389 40 20,685 99,476

8004 LCM Other 24,216 38 11,924 36,940

Combined Total 503,070 542 188,304 860,726

Percent of Total Crashes 8.7% 1.6% 6.7% 11.8%

201

Table A-4: Target Population for PAEB

Crash Scenarios Crashes Fatalities MAIS 1-5 Injuries PDOVs

300 1V2Ped RD, Forward

Impact

60,322 3,264 57,480 1,836

309 1V2Ped, Other 306 26 264 0

350 2+V2Ped 511 259 452 0

400 1V2Cyc RD, Forward

Impact

50,094 531 45,529 4,910

409 1V2Cyc,

Other/Unspecified

175 4 172 0

450 2+V2Cyc 234 23 169 239

Combined Total 111,641 4,106 104,066 6,985

Percent of Total Crashes 1.9% 12.3% 3.7% 0.1%

Table A-5: Target Population for RAB/RvAB/RCTA Technologies

Crash Scenarios Crashes Fatalities MAIS 1-5 Injuries PDOVs

302 1V2Ped, Backup 2,811 44 2,590 88

402 1V2Cyc, Backup 439 3 407 48

602 1V2ParkedV, Backup 41,957 2 5,293 40,389

802 1V2Fixed Object,

Backup

1,824 2 217 1,732

6000 Backing Up to

Vehicle/Object

101,503 23 26,761 189,059

Combined Total 148,533 74 35,268 231,317

Percent of Total Crashes 2.6% 0.2% 1.3% 3.2%

202

Table A-6: Mapping of Crash Scenarios with Safety Systems

1 2 3 4 5

Crash Scenarios FCW/CIB/DBS LDW/LKA/LCA BSD/BSI/LCM PAEB RAB/RvAB/RTA

100 1V Rollover 1st Event

●

150 2+V Rollover 1st Event

●

200 1V Jackknife 1st Event

250 2+V Jackknife 1st Event

300 1V2Pedestrian Roadway

Departure, Forward Impact

●

302 1V2 Pedestrian, Backup

●

309 1V2 Pedestrian, Specifics

Other/Unknown

●

350 2+V2 Pedestrian

●

400 1V2Cyclist Roadway

Departure, Forward Impact

●

402 1V2Cyclist, Backup

●

409 1V2Cyclist, Specifics

Other/Unknown

●

450 2+V2Cyclist

●

500 1V2Animal Roadway

Departure, Avoid Animal

502 1V2Animal, Backup

509 1V2Animal, Specifics

Other/Unknown

550 2+V2Animal

600 1V2Parked Vehicle

Roadway

Departure, Forward Impac

602 1V2Parked Vehicle,

Backup

●

1 2 3 4 5

Crash Scenarios FCW/CIB/DBS LDW/LKA/LCA BSD/BSI/LCM PAEB RAB/RvAB/RTA

609 1V2Parked Vehicle,

Specifics Other/Unknown

650 2+V2Parked Vehicle

700 1V2Other Non-Fixed

Object Roadway Departure,

Forward Impac

203

701 1V2Other Non-Fixed

Object Roadway Departure,

Traction Loss

702 1V2Other Non-Fixed

Object,

Backup

709 1V2Other Non-Fixed

Object,

Othe

750 2+V2Other Non-Fixed

800 1V2Fixed Object Roadway

Departure, Forward Impact

801 1V2Fixed Object Roadway

Departure, Traction Loss

802 1V2Fixed Object, Backup

●

809 1V2Fixed Object, Other

850 2+V2Fixed Object

1000 1V, Roadway Departure

●

1001 1V RD, Traction Loss

1002 1V RD, Avoid

Vehicle/Pedestrian/Animal

1003 1V Forward Impact, Ped

or Animal

1 2 3 4 5

Crash Scenarios FCW/CIB/DBS LDW/LKA/LCA BSD/BSI/LCM PAEB RAB/RvAB/RTA

1004 1V Forward Impact, End

Departure

1005 1V Forward Impact,

Specifics

Other/Unknown

1009 1V Other/No Impact

1050 2+V, Roadway Departure

●

1100 1V Cross

Centerline/Median

●

1150 2+V Cross

Centerline/Median*

●

2000 Rear-End, Lead Vehicle

Stopped

●

2001 Rear-End, LV Slower ●

2002 Rear-End, LV Decelerated ●

204

2003 Rear-End, Other In-lane

Vehicle Higher Speed

●

2009 Rear-End, Specifics

Other/Unknown

●

2101 Same Trafficway Same

Direction Forward Impact, Loss

Control

2102 Rear-End Possible, Same

Trafficway Same Direction

Forward

Impact, Avoid Vehicle

2103 Same Trafficway Same

Direction Forward Impact,

Avoid Objects

2109 Rear-End Possible, Same

Trafficway Same Direction

Forward Impact, Specifics

Other/Unknown

1 2 3 4 5

Crash Scenarios FCW/CIB/DBS LDW/LKA/LCA BSD/BSI/LCM PAEB RAB/RvAB/RTA

2200 Same Trafficway Same

Direction, Angle-Sideswipe

2300 Rear-End Possible, Other

In-

lane Vehicle Stoppe

●

2301 Rear-End Possible, Other

In-

lane Vehicle Slower

●

2302 Rear-End Possible, Other

In- lane Vehicle Decelerated

●

3000 Same Trafficway Opposite

Direction, Head-On

●

3001 Same Trafficway Opposite

Direction Forward Impact,

Traction Loss

3002 Same Trafficway Opposite

Direction Forward Impact,

Avoid Vehicle

205

3003 Same Trafficway Opposite

Direction Forward Impact,

Avoid

3009 Same Trafficway Opposite

Direction Forward Impact,

Other

●

3100 Same Trafficway Opposite

Direction, Angle Sideswipe

●

3200 Head-On Possible, Other

Vehicle Encroaching Opposite

Direction

●

1 2 3 4 5

Crash Scenarios FCW/CIB/DBS LDW/LKA/LCA BSD/BSI/LCM PAEB RAB/RvAB/RTA

4000 Change Trafficway

Vehicle Turning, Turn Across

Path, Initial Opposite Direction

4001 Change Trafficway

Vehicle Turning, Turn Across

Path, Initial

Same Direction

4009 Change Trafficway

Vehicle Turing, Turn Across

Path, Specifics

Other/Unknown

4100 Change Trafficway

Vehicle Turning, Turn Into

Path, Into Same

Direction

4101 Change Trafficway

Vehicle Turning, Turn Into

Path, Into Opposite Direction

4109 Change Trafficway

Vehicle Turning, Turn Into

Path, Specifics Other/Unknown

5000 Intersect Paths, Straight

Across

Path

5009 Intersect Paths, Straight

Path,

Specifics, Specifics

Other/Unknown

6000 Backing Up to

Vehicle/Object

●

7000 1V Negotiating a Curve

7050 2+V Negotiating a Curve

206

8000 Lane Change/Merge

Before Rear-End

●

1 2 3 4 5

Crash Scenarios FCW/CIB/DBS LDW/LKA/LCA BSD/BSI/LCM PAEB RAB/RvAB/RTA

8001 Lane Change/Merge in

Same Trafficway Same

Direction Forward

Impac

●

8002 Lane Change/Merge in

Same Trafficway Same

Direction Angle Sideswipe

●

8003 Lane Change/Merge in

Change Trafficway Vehicle

Turing Initial Same Direction

●

8004 Lane Change/Merge Other

●

9000 Equipment Failure

9020 Loss of Control Due to

Tire/Engine/Poor Road

9030 2+V, Left/Right Turn,

Unspecified

9040 2+V U-Turn

9050 2+V Backing to Moving

Vehicle

9060 2+V No Impact

9070 2+V Other

9999 2+V Unknown

207

Appendix B. Questions Asked Throughout This Notice

III. ADAS Performance Testing Program

(1) Should the Agency award credit to vehicles equipped with LDW systems that provide a

passing alert, regardless of the alert type? Why or why not? Are there any LDW alert

modalities, such as visual-only warnings, that the Agency should not consider

acceptable when determining whether a vehicle meets NCAP’s performance test

criteria? If so, why? Should the Agency consider only certain alert modalities (such as

haptic warnings) because they are more effective at re-engaging the driver and/or have

higher consumer acceptance? If so, which one(s) and why?

(2) If NHTSA were to adopt the lane keeping assist test methods from the Euro NCAP LSS

protocol for the Agency’s LKS test procedure, should the LDW test procedure be

removed from its NCAP program entirely and an LDW requirement be integrated into

the LKS test procedure instead? Why or why not? For systems that have both LDW

and LKS capabilities, the Agency would simply turn off LKS to conduct the LDW test if

both systems are to be assessed separately. What tolerances would be appropriate for

each test, and why?

(3) LKS system designs provide steering and/or braking to address lane departures (e.g.,

when a driver is distracted). To help re-engage a driver, should the Agency specify that

an LDW alert must be provided when the LKS is activated? Why or why not?

(4) Do commenters agree that the Agency should remove the Botts’ Dots test scenario from

the current LDW test procedure since this lane marking type is being removed from use

in California? If not, why?

(5) Is the Euro NCAP maximum excursion limit of 0.3 m (1.0 ft.) over the lane marking (as

defined with respect to the inside edge of the lane line) for LKS technology acceptable,

or should the limit be reduced to account for crashes occurring on roads with limited

208

shoulder width? If the tolerance should be reduced, what tolerance would be appropriate

and why? Should this tolerance be adopted for LDW in addition to LKS? Why or why