ID: nht92-5.39

DATE: June 29, 1992

FROM: Paul Jackson Rice -- Chief Counsel, NHTSA

TO: Steven Henderson -- Department of Psychology, McGill University, Montreal, Canada

TITLE: None

ATTACHMT: Attached to letter (fax) dated 6/3/92 from Steven Henderson to Michael Perel (OCC 7383)

TEXT:

Michael Perel of this agency has asked this Office to reply to your FAX of June 3, 1992, in which you have asked for a legal opinion with respect to the relationship of your motorcycle headlamp warning-device to S5.6 of Federal Motor Vehicle Safety Standard No. 108.

As you have described the device, it "flickers the headlight, tail light, and signal lights at a rate of 10 flashes per second whenever the horn button is pressed." If the horn button is pushed while a turn signal is operating, "the headlight flickers at a rate of 10 Hz, and only the actuated signal light gives out 2 Hz bursts of 10 Hz flicker." The device "is also intended for use with automobile signal lights (but not automobile headlights)."

One of the basic performance requirements of Standard No. 108 is that lighting equipment in use be steady-burning, with the exceptions provided in S5.5.10. One of these exceptions (S5.5.10(c)) allows either the upper beam, or the lower beam, of a motorcycle headlamp to be wired to modulate from a higher intensity to a lower intensity, in accordance with S5.6. Your letter does not state whether the "flicker" is between the upper and lower beams, or between intensity levels within a single beam so that we do not know whether the modulation is in accordance with S5.6. However, the flash rate of 10 per second is substantially more than the maximum of 280 cycles per minute specified by S5.6.1(a), so, that overall, the device would not comply with the exception set forth in S5.5.10(c).

Furthermore, the device would affect compliance of the taillamps and turn signal lamps with Standard No. 108. The taillamps would no longer be steady-burning, as required by S5.5.10(d). It would appear that the turn signal rate would also cease to comply with the flash rate of 60-120 per minute specified by SAE requirements incorporated by reference in Standard No. 108.

For the reasons stated above, the device would be impermissible for installation as original motor vehicle equipment. The situation is much the same in the aftermarket. Under 15 U.S.C. 1397(a)(2)(A), as it applies to your device, no manufacturer, distributor, dealer, or motor vehicle-repair business may modify a motor vehicle in a manner that renders it noncompliant with Standard No. 108. You will note that the statute does not preclude the owner from such modifications. Thus, if a motorcycle owner can install the device, there would be no violation of Federal Law. In this event, the legality of its use would be determinable under the laws of the individual American states, many of which follow Standard No. 108.

You have also asked our assistance "in advancing the examination of a U.S. patent application" for the device in accordance with 37 CFR 1.102. Advancements may occur upon application wherein the invention is "deemed of peculiar importance to some branch of the public service and the head of some department of the Government requests immediate action for that reason." Because the safety benefits of the device are speculative and unproven, and because the device is contrary to the requirements of Standard No. 108, I believe that it would be improper for this agency to intervene in the manner you have requested.

ID: 18558.ztv

Mr. Jim Ashdown
E & D Manager
Magnetti Marelli UK Ltd.
Lighting Division
Walkmill Lane, Cannock
Staffordshire WS11 3LP
England

FAX 9-0111543466325

Dear Mr. Ashdown:

This is in reply to your fax of August 6, 1998, to Taylor Vinson of this Office. Because of the need to coordinate interpretations of Federal motor vehicle safety standards with other Offices of this agency, we were unable to respond by "return fax" before now.

You have two questions and they relate to Federal Motor Vehicle Safety Standard No. 108. You ask for confirmation that "paragraph S.7.8.5.2.C is the correct section for tamper proof V HAD systems." We confirm your understanding. Paragraph S7.8.5.2(c) requires that headlamps with VHADs manufactured for use on motor vehicles manufactured on or after September 1, 1998, shall be manufactured with their calibration permanently fixed by the headlamp manufacturer.

Your second questions is whether there is "a requirement now for optical centre markings on rearlamps similar to S.7.8.5.3.f on headlamps." There is no such requirement. Paragraph S7.8.5.3(f) requires such markings only on headlamps. There is no requirement in Standard No. 108 that the optical center be marked on any lamp other than a headlamp.

Sincerely,
Frank Seales, Jr.
Chief Counsel
ref:108
d.9/22/98

ID: nht79-2.46

DATE: 01/19/79

FROM: AUTHOR UNAVAILABLE; M. M. Finkelstein; NHTSA

TO: Stanley Electric Company, Ltd.

TITLE: FMVSS INTERPRETATION

TEXT: This is in reply to your letter of June 14, 1978, to Mr. E. T. Driver, requesting an interpretation of Federal Motor Vehicle Safety Standard (FMVSS) No. 108, Lamps, Reflective Devices and Associated Equipment. Your letter depicted the use of two headlamps, both round and rectangular, mounted side by side, and also one above the other.

Table IV of FMVSS No. 108 requires that if two headlamps are mounted on a motorcycle they must be symmetrically disposed about the vertical centerline. The same requirement applies to taillamps, stoplamps, and reflex reflectors. This has been interpreted previously, in the case of reflex reflectors, to mean that if two are used they may be mounted only side by side. Four headlamp systems are not permitted on motorcycles. These are specified in the new paragraph S4.1.1.34 in the amendment published in the Federal Register July 27, 1978. A copy of the new amendment is enclosed, along with copies of previous notices that constitute FMVSS No. 108. However, if two headlamps are mounted on a motorcycle, they still must be mounted side by side and equidistant from a vertical centerline of the body of the motorcycle.

This Agency is currently considering an amendment to the standard that will, if approved, permit the positioning of headlamps and reflectors one above the other when two are mounted on a motorcycle. The necessary rulemaking procedures, if initiated, would take several months and there is no certainty that the contemplated amendment would be issued.

We welcome your further comments and questions.

ID: 2953o

Mr. George Ziolo
DOT Paperwork Processor
234l7 Everett Place
Ramona, CA 92065

Dear Mr. Ziolo:

This is in reply to your letter of April 20, 1988, asking about the acceptability under Safety Standard No. l08 of modifying imported vehicles so that they are equipped with two Type 2D1 and two Type 1C1 headlamps. You have been informed by the agency's Office of Vehicle Safety Compliance that this is impermissible "because they are 'nonconforming' 'headlight systems'." You disagree because you believe that the minimum requirements of the Standard are satisfied by the Type 2D1 lamps, and that "S4.4 appears to permit such a combination."

Paragraph S4.4 is not applicable to the situation you present as it refers to combinations of lamps serving different functions; in your discussion, the lamps serve the identical function of headlighting. Given the fact that the Type 2D1 sealed beam 7" diameter headlamps fulfill the headlighting requirements of the Standard, your question must be viewed as whether a supplement to the headlighting system is permissible under Standard No. l08.

Paragraph S4.l.3 of Standard No. l08 permits the addition of nonrequired lighting equipment provided it does not impair the effectiveness of the equipment that the standard requires. The two Type 1C1 5 3/4" diameter sealed beam lamps in a four lamp headlighting system form the major portion of an upper beam headlighting system. The two Type 2D1 lamps in a two lamp headlighting system form the whole of an upper beam headlighting system. Thus, a vehicle furnished with the systems you posit would be equipped with more than one upper beam headlighting system. The Type 2D1 system must be designed to conform to the photometric requirements of SAE Standard J579c DEC79 "Sealed Beam Headlamp Units for Motor Vehicles." The SAE Standard establishes at two test points, H-V and 4 D-V, maximum allowable candela of 75,000 and 5,000 respectively for each Type 2D1 headlamp. This means that the maximum allowable candela for headlighting systems at these test points is 150,000 and l0,000 candela. The Type 1C1 headlamps will also be designed to conform to SAE J579c. Corresponding maxima at test points H-V and 4 D-V for Type 1C1 systems are 60,000 and 5,000. Thus, a vehicle equipped with the lamps you have described could emit a total of 270,000 candela at test point H-V (when only 150,000 is permitted), and 20,000 at 4D-V (when only l0,000 is allowable).

Agency research has shown that candela readings in excess of 150,000 greatly increase the potential for glare with little increase in seeing ability. This glare would be visible both to the driver of an oncoming car, and the driver of the modified vehicle itself through creation of a "veiling" glare. The addition of the Type 1C1 headlamps would therefore impair the effectiveness of the Type 2D1 headlighting system, and is forbidden by S4.1.3.

We appreciate your interest in safety.

Sincerely,

Erika Z. Jones Chief Counsel

ref:108 d:9/12/88

ID: Google -- compiled response to 12 Nov 15 interp request -- 4 Feb 16 final

Chris Urmson

Director, Self-Driving Car Project

Google, Inc.

1600 Amphitheatre Parkway

Mountain View, CA 94043

Dear Dr. Urmson:

This responds to your November 12, 2015 letter[1] requesting that the National Highway Traffic Safety Administration (NHTSA) interpret a number of provisions in the Federal Motor Vehicle Safety Standards (FMVSSs) as they apply to Google’s described design for motor vehicles that it is in the process of developing and testing.  According to Google, those self-driving vehicles (SDVs) are “fully autonomous motor vehicles, i.e., vehicles whose operations are controlled exclusively by a Self-Driving System (SDS).”  The SDS is an artificial-intelligence (AI) “driver,” which is a computer designed into the motor vehicle itself that controls all aspects of driving by perceiving its environment and responding to it.[2]  Thus, Google believes that the vehicles “have no need for a human driver.” 

In this response, NHTSA addresses each of Google’s requests for interpretation, and grants several of them.   In some instances, the issues presented simply are not susceptible to interpretation and must be resolved through rulemaking or other regulatory means.  NHTSA believes that many of these issues may be resolved on an interim basis through well-supported exemption petition(s), and invites Google to file such petitions.  In other instances, if Google is able to provide sufficient additional information and evidence, it may be possible to resolve open issues through interpretations.

INTRODUCTION

Google describes its vehicles as having what NHTSA’s May 2013 Preliminary Statement of Policy Concerning Automated Vehicles calls Level 4 Full Self-Driving Automation.  According to that Statement, a Level 4 vehicle

is designed to perform all safety-critical driving functions and monitor roadway conditions for an entire trip. Such a design anticipates that the driver will provide destination or navigation input, but is not expected to be available for control at any time during the trip. This includes both occupied and unoccupied vehicles. By design, safe operation rests solely on the automated vehicle system.

Google is asking for interpretations to determine how it would certify its SDV to the FMVSS.  In essence, Google seeks to produce a vehicle that contains L4 automated driving capabilities, and removes conventional driver controls and interfaces (like a steering wheel, throttle pedal, and brake pedal, among many other things).  Given that the SDS controls all aspects of driving, and given Google’s belief that the SDS consistently will make the optimal decisions for the SDV occupants’ safety (as well as for pedestrians and other road users), the company expresses concern that providing human occupants of the vehicle with mechanisms to control things like steering, acceleration, braking, or turn signals, or providing human occupants with information about vehicle operation controlled entirely by the SDS, could be detrimental to safety because the human occupants could attempt to override the SDS’s decisions.  While the L4 automation is the impetus behind these design decisions, it is Google’s design decisions that create the uncertainty over how to apply the FMVSS to Google’s proposed vehicle. 

Google’s design choices in its proposed approach to the SDV raise a number of novel issues in applying the FMVSSs.  Those standards were drafted at a time when it was reasonable to assume that all motor vehicles would have a steering wheel, accelerator pedal, and brake pedal, almost always located at the front left seating position, and that all vehicles would be operated by a human driver.  Accordingly, many of the FMVSSs require that a vehicle device or basic feature be located at or near the driver or the driver’s seating position.  For vehicles with an AI driver that also preclude any occupant from assuming the driving task, these assumptions about a human driver and vehicle controls do not hold.  As self-driving technology moves beyond what was envisioned at the time when standards were issued, NHTSA may not be able to use the same kinds of test procedures for determining compliance.  And since the Safety Act creates a self-certification system for compliance, NHTSA’s verification of a manufacturer’s compliance (and thus, the agency’s ability to enforce against non-compliance) is based on our established test procedures.

In order to determine where to place vehicle devices and features, or whether to provide them at all, Google has asked who or what is to be considered the “driver” and which seating position is considered to be the “driver’s seating position” in its SDV.  49 CFR 571.3 defines “driver” as “the occupant of a motor vehicle seated immediately behind the steering control system.”  Because Google’s SDV design purposely does not have any mechanism by which human occupants could steer or otherwise “drive” the vehicle, it would be difficult in several instances to determine who the “driver” would be in its SDV, and thus to determine how to certify its motor vehicle design to certain FMVSS provisions that reference that “driver.”  To attempt to solve this challenge, Google has offered different interpretations of “driver” or “driver’s seating position” for NHTSA’s consideration, varying with the specific requirement or circumstances, and requests that we confirm its interpretations.

Specifically, Google suggests two potential interpretations of “driver” and one potential interpretation for “driver’s position” or “driver’s designated seating position” in the context of its described motor vehicle design:

• NHTSA could interpret the term “driver” as meaningless for purposes of Google’s SDV, since there is no human driver, and consider FMVSS provisions that refer to a driver as simply inapplicable to Google’s vehicle design;
• NHTSA could interpret “driver” and “operator” as referring to the SDS; or
• NHTSA could interpret “driver’s position” or “driver’s designated seating position” as referring to the left front outboard seating position, regardless of whether the occupant of that position is able to control the vehicle’s operation or movements.

Google then applies these suggested interpretive approaches to a number of FMVSS provisions in order to justify being able to certify its intended motor vehicle design as compliant with the FMVSSs, without having to change the design in ways that Google finds problematic. 

We will address each of Google’s suggested interpretations in turn.  The critical point of NHTSA’s responses for many of the requested interpretations is that defining the driver as the SDS (or the driver’s position as the left front position) does not end the inquiry or determine the result.  Once the SDS is deemed to be the driver for purposes of a particular standard or test, the next question is whether and how Google could certify that the SDS meets a standard developed and designed to apply to a vehicle with a human driver.  Related, in order for NHTSA to interpret a standard as allowing certification of compliance by a vehicle manufacturer, NHTSA must first have a test procedure or other means of verifying such compliance.

While some of Google’s requested interpretations may be permissible given the facts presented here, we wish to make clear that many of the other requests present policy issues beyond the scope and limitations of interpretations and thus will need to be addressed using other regulatory tools or approaches.  NHTSA further notes that in a number of instances (in particular, several included in Table B), it may be possible for Google to show that certain standards are unnecessary for a particular vehicle design.  To date, however, Google has not made such a showing.  We note that these interpretations are confined to the specific facts and circumstances set forth in Google’s letter, and that they do not apply to other facts and circumstances.  We also emphasize that the interpretations NHTSA is issuing are subject to change or revocation if new or different facts or information comes to light.

DISCUSSION

NHTSA is authorized by the National Traffic and Motor Vehicle Safety Act (“Safety Act,” 49 U.S.C. Chapter 301) to issue FMVSSs that set performance requirements for new motor vehicles and new items of motor vehicle equipment.  Under the Safety Act, NHTSA does not provide approvals of motor vehicles or motor vehicle equipment and does not make determinations as to whether a product conforms to the FMVSSs outside of a NHTSA compliance test.  Instead, the Safety Act requires manufacturers to self-certify that their products conform to all applicable FMVSSs that are in effect on the date of manufacture.  Manufacturers are also responsible for ensuring that their products are free of safety-related defects.  NHTSA enforces compliance with the FMVSSs by testing vehicles and regulated equipment.  NHTSA also investigates safety-related defects and conducts related enforcement and recall actions. 

As a foundational starting point for the interpretations below, NHTSA will interpret “driver” in the context of Google’s described motor vehicle design as referring to the SDS, and not to any of the vehicle occupants.  We agree with Google its SDV will not have a “driver” in the traditional sense that vehicles have had drivers during the last more than one hundred years.  The trend toward computer-driven vehicles began with such features as antilock brakes, electronic stability control, and air bags, continuing today with automatic emergency braking, forward crash warning, and lane departure warnings, and continuing on toward vehicles with Google’s SDV and potentially beyond.  No human occupant of the SDV could meet the definition of “driver” in Section 571.3 given Google’s described motor vehicle design – even if it were possible for a human occupant to determine the location of Google’s steering control system, and sit “immediately behind” it, that human occupant would not be capable of actually driving the vehicle as described by Google.  If no human occupant of the vehicle can actually drive the vehicle, it is more reasonable to identify the “driver” as whatever (as opposed to whoever) is doing the driving.  In this instance, an item of motor vehicle equipment, the SDS, is actually driving the vehicle. 

NHTSA will consider initiating rulemaking to address whether the definition of “driver” in Section 571.3 should be updated in response to changing circumstances, as contemplated by Executive Order 12866, Regulatory Planning and Review.  Section 5, Reducing Regulations, of that Order provides that each federal regulatory agency will

[i]n order … to determine whether regulations promulgated by the executive branch of the Federal Government have become unjustified or unnecessary as a result of changed circumstances …  periodically review its existing significant regulations to determine whether any such regulations should be modified or eliminated so as to make the agency’s regulatory program more effective in achieving the regulatory objectives …

NHTSA recognizes that it can take substantial periods of time to develop some rulemaking proposals and final rules, including time spent obtaining review of those proposals and seeking and analyzing public comments.  NHTSA further understands that the time it takes to conduct rulemakings may, in some instances, make such proceedings ill-suited as first-line regulatory mechanisms to address rapidly-evolving vehicle technologies.  That said, there are limits to the result the agency may reach in an interpretation, even if it believes that result might be sound policy.  An interpretation describes an agency’s view of the meaning of an existing statute or regulation.  It can make sense of the overall legal framework and provide clarity for regulated entities and the public.  For example, an interpretation may clarify a statutory or regulatory term or provide crisper and more detailed lines than the regulation or statute being interpreted.  An interpretation may not, however, make a substantive change to the statutory or regulatory regime or to the clear language of a provision.  In particular, an interpretation may not adopt a new position that is irreconcilable with or repudiates existing statutory or regulatory provisions.   

In many instances, interpreting the term “driver” in a manner that Google has requested does not necessarily change the requirements of the regulation or otherwise fully resolve the issue Google seeks to address. Because the interpretations provided by this letter do not fully resolve all of the issues Google has raised, Google may wish to explore the interim step of seeking exemptions.  Exemptions are available under 49 U.S.C. 30114 and 49 CFR Part 555 for manufacturers able to demonstrate that features of their products provide equivalent levels of safety to those required by the FMVSS.  

1. “Priority Interpretive Issues”

Google requested NHTSA’s interpretation of several “priority interpretive issues” related to the absence of a human driver.  Google also provided two tables, Attachments A and B, listing FMVSS provisions for which Google requested that NHTSA interpret the “driver” or “operator” to be the SDS (provisions in Attachment A) and other provisions for which Google requested that NHTSA interpret “driver” to be the human occupant seated in the left front designated seating position (primarily provisions in Attachment B).  We address these interpretation requests in the order presented by Google’s November 12 letter.

FMVSS No. 135, “Light Vehicle Brake Systems”

Deeming the SDS to be the “driver” of a motor vehicle does not excuse that vehicle from compliance with the brake requirements of applicable standards.  FMVSS No. 135 contains requirements for service brakes and associated parking brake systems.  Among these requirements is S5.3.1, which states that service brakes “shall be activated by means of a foot control,” and also that “control of the parking brake shall be independent of the service brake control, and may be either a hand or foot control.”  Google’s described motor vehicle design does not include hand or foot controls for either the service brakes or the parking brake.  Google argues that because the SDS will control all aspects of braking, it would not be necessary or beneficial for safety for a human occupant to be able to brake the vehicle.  Google therefore requests that NHTSA interpret these provisions regarding the activation or control of braking systems to be inapplicable to its described motor vehicle design.  Similarly, Google requests that NHTSA interpret S6.5.1 of FMVSS No. 135 to allow the service brake system performance requirements to be met if the SDS activates the service brakes, rather than “solely by use of the service brake control” as the provision directs. 

We agree that Google’s SDS may be deemed to be the driver for purposes of compliance with these provisions.  Given that there will be no foot (or even hand) control to be activated – indeed, given that the SDS will have neither feet nor hands to activate brakes – we understand that Google’s described vehicle design would not comply with S5.3.1 as written.  We also understand Google’s assertions that the SDS will be able to activate the brakes electronically such that its vehicle will “stop [] safely and in accordance with all performance requirements of FMVSS No. 135.”  The fact that the SDS may be programmed to perform the tests enumerated in FMVSS No. 135 does not, however, overcome the plain language of S5.3.1.  NHTSA would need to commence a rulemaking to consider how FMVSS No. 135 might be amended in response to “changed circumstances” in order to ensure that automated vehicle designs like Google’s, i.e., ones that control all braking through an AI driver and do not provide brake controls to vehicle occupants, have a way to comply with the standard.  Such a rulemaking would also consider S6.5.1 and any other provisions that implicate the potential use of a foot-actuated service brake control.  In the interim, Google may wish to consider petitioning the agency for an exemption from these provisions.[3]

FMVSS No. 101, “Controls and Displays”

FMVSS No. 101 contains requirements for location, identification, color, and illumination of motor vehicle controls, telltales, and indicators.  S5.1.1 requires the controls listed in Tables 1 and 2 of the standard to be “located so that they are operable by the [belted] driver.”  S5.1.2 requires the telltales and indicators in those Tables and their identification to be “located so that, when activated, they are visible to a [belted] driver.”  Google states that its motor vehicle will not have any of the controls listed in Tables 1 and 2 because there will not be a human driver in its vehicle.  Further, Google believes that allowing the human occupants to affect the operation of, for example, lamps or turn signals otherwise controlled by the SDS could be detrimental to safety.  Google therefore asks that NHTSA interpret S5.1.1 to be inapplicable to its motor vehicle design, or that NHTSA interpret Google’s SDS to be the “driver” for the purposes of S5.1.1.  For S5.1.2, Google says it will equip its vehicle’s occupant compartment with the telltales and indicators required by other FMVSSs (like the telltales/indicators for malfunctions of the brake system, electronic stability control system, or tire pressure monitoring system).  For purposes of placement of those telltales and indicators, Google requests that NHTSA interpret the “driver position” to be the left front seating position.

Because we interpret “driver” as referring to the SDS, we agree with Google that the controls listed in Tables 1 and 2 may simply be operable by the SDS and need not be located so that they are available to any of the human occupants of the motor vehicle.  Similarly, telltales and indicators must also be “visible” to the SDS.  For purposes of both S5.1.1 and S5.1.2, we interpret the SDS to be the “driver.”  However, we have no defined way at this time of verifying Google’s compliance with this interpretation of these requirements.  Thus, if Google certified its compliance with these provisions consistent with this interpretation, NHTSA would be unable to conduct confirmatory testing to satisfy ourselves that the Google vehicle is compliant.  Therefore, unless and until NHTSA has a standard and testing procedures to confirm compliance with S5.1.1 and S5.1.2, or a standard providing equal or greater safety, it cannot conclude that Google’s SDV is compliant with those requirements.  In order to determine what “operable by” and “visible to” the SDS mean, and to establish procedures for testing compliance with those requirements using its existing regulatory tools, NHTSA would be required to conduct a rulemaking.

In the interim, Google may wish to petition the agency for an exemption from these provisions.[4]  Additionally, we agree with Google that it could be beneficial to vehicle occupants to be aware of certain aspects of vehicle status through telltales and indicators while they are traveling in the vehicle.  Given the historical status of the left front seating position as the location of the human in charge of vehicle operation, we agree that Google may voluntarily locate telltales and indicators so that they are visible to that position.  However, if it is foreseeable that a different seating position is more likely to be occupied, it may make more sense for the telltales and indicators to be located so that they are visible in another position.

FMVSS No. 108, “Lamps, Reflective Devices, and Associated Equipment”

FMVSS No. 108 contains requirements for original and replacement lamps, reflective devices, and associated equipment.  Google states that it “recognizes that all vehicles must comply with all of the performance requirements of FMVSS No. 108 with respect to external lamps and reflectors,” but asks that NHTSA interpret S6.6.1 of FMVSS No. 108 to allow certain internal controls “to be excluded from the occupant compartment, as long as their functions are performed by the vehicle autonomously.”  S6.6.1 requires all vehicles covered by FMVSS No. 108 to be “equipped with a turn signal operating unit, a turn signal flasher, a turn signal pilot indicator, a headlamp beam switching device, and an upper beam headlamp indicator meeting the requirements of S9.” 

With respect to turn signal operating units, Google requests that NHTSA interpret the term “operator” in the S4 definition of “turn signal operating unit” to mean the SDS.  That definition states that the unit is “an operating unit that is part of a turn signal system by which the operator of a vehicle causes the signal unit to function.”  Google notes that, in its vehicle, the turn signals will be operated and controlled by the SDS.  Google’s letter also addresses S9.1.1, which requires that the “turn signal operating unit…must be self-canceling by steering wheel rotation and capable of cancellation by a manually operated control.”  Google requests that NHTSA interpret this provision to be met by Google’s SDS since there would not be any steering wheel or conventional turn signal stalk.  Google also argues that providing either a steering wheel or the ability to control turn signal operation to the vehicle occupants could be detrimental to safety.  Google further asks that NHTSA “expressly acknowledg[e]” that a steering wheel is not required by the FMVSS.  The supplemental information Google provided on January 11, 2016 stated that the SDV’s turn signal operating unit self-cancels based on the position of the steering rack (which is controlled by the SDS), “which is what happens in a conventional vehicle by virtue of the link between the steering rack, steering column, and steering wheel,” and that Google therefore believes that the SDV complies with S9.1.1.

With respect to the S9.4 requirement for headlamp beam switching devices, which provides that “[e]ach vehicle must have a means of switching between lower and upper beams designed and located so that it may be operated conveniently by a simple movement of the driver’s hand or foot,” Google again states that the SDS will control headlamp beam switching “fully and appropriately” and that human occupant control over such functions could be detrimental to safety.

For purposes of these provisions of FMVSS No. 108, we interpret the SDS to be the “driver.”  We also agree that the “operator” of a turn signal operating unit is the SDS, because NHTSA has typically used “driver” and “operator” interchangeably in its interpretations over time.  We further agree that a steering wheel is not expressly required by any FMVSS.  We agree with Google that, as described, the SDV appears to be designed to self-cancel the turn signal operating unit as NHTSA would have intended, had vehicles without steering wheels been available when FMVSS No. 108 was developed.  Similar to the provisions of FMVSS No. 135, however, the fact that the SDS may be programmed to cancel the turn signal and switch headlamp beams does not overcome the (other) plain language of S9.1.1 and S9.4.  Even if we agree that Google’s self-cancelation for the turn signal operating unit is consistent with our intent in S9.1.1., that provision further expressly requires that the turn signal be capable of cancellation by a “manually operated control.” We understand Google’s argument that in the case of the SDV, manual cancelation of the turn signal operating unit by a human occupant could interfere with decisions already made by the SDS in ways that could affect safety.  However, this is an issue that cannot be addressed through interpretation alone.  Similarly, S9.4 specifically requires operation of headlamp switches by movement of the “driver’s hand or foot”.[5]  Under existing test procedures, we cannot verify Google’s compliance with these express requirements. 

Unless and until NHTSA has a standard and testing procedures to confirm compliance with these provisions, it cannot interpret Google’s SDV as compliant with these standards and requirements.  In order to determine what requirements would be appropriate, and to establish procedures for testing compliance with those standards, using its existing regulatory tools, NHTSA would be required to conduct a rulemaking.  NHTSA would need to consider how it could propose to amend FMVSS No. 108 in order to ensure that automated vehicle designs like Google’s, i.e., those that control all lighting and signals through the AI driver and do not provide lighting or signal controls to vehicle occupants, have a way to comply with the standard.  Meanwhile, Google may wish to petition the agency for an exemption from these provisions.  An exemption petition could seek to demonstrate that Google’s SDV would provide an equivalent level of safety to that provided by compliance with the FMVSS.[6] 

1. Interpretive Issues regarding “Miscellaneous Provisions in Other FMVSS”

Google also requested interpretation of several other “miscellaneous provisions in other FMVSS” relating to the absence of a human driver.

FMVSS No. 111, “Rear Visibility”

FMVSS No. 111 contains requirements for rear visibility devices and systems, requiring that vehicles have external and internal rear view mirrors to provide the driver with certain fields of view around and behind the vehicle.  FMVSS No. 111 also requires that vehicles display a rearview image (of a specified area of certain dimensions behind the vehicle) to the vehicle operator.  Google asks that NHTSA interpret these provisions as requiring that the specified view be provided to the SDS, rather than to the human occupant.  Google further requests that “the vehicle would be deemed compliant if the SDS receives sensor input at least equivalent to the images a driver would be able to view through mirrors and a rear visibility system meeting the field of view and other performance requirements of the standard.”

As above, because we are interpreting “driver” and “operator” in this instance to refer to the SDS, we agree that the information required by the provisions of FMVSS No. 111 that must be provided to the “driver” or “operator” may be provided to the SDS.  Here again, we have no defined way at this time of verifying Google’s compliance with this interpretation of those requirements.  Thus, if Google certified its compliance with these provisions consistent with this interpretation, NHTSA would be unable to conduct confirmatory testing to satisfy ourselves that such certification was valid. 

Therefore, unless and until NHTSA develops and adopts appropriate performance criteria and test procedures for evaluating whether the sensor input received by the SDS provides enough information to ensure that the SDS is as well informed by its sensors of the conditions behind and around it as a human driver of a conventional vehicle that meets the existing requirements of FMVSS No. 111 (or a standard providing equal or greater safety), it cannot interpret Google’s SDV as compliant with these standards and requirements.  This would need to be undertaken through rulemaking.  NHTSA may also consider as part of such a rulemaking whether there is benefit to conveying this information to human occupants in the case of vehicle designs like the Google SDV.  Google may wish to petition the agency for an exemption from these provisions, in which it could seek to demonstrate that its SDV would provide an equivalent level of safety to that provided by compliance with the FMVSS.[7] 

FMVSS No. 114, “Theft Protection and Rollaway Prevention”

FMVSS No. 114 contains requirements intended to reduce the incidence of crashes resulting from theft and accidental rollaway of vehicles.  Among these requirements is S5.3, which requires vehicles with an automatic transmission that includes a “park” position to have a system that “requires the service brake to be depressed before the transmission can be shifted out of ‘park.’”  Google states that its vehicle will not have any brake pedal, and that its SDS “will determine the appropriate transmission position and will not select a position other than park unless the service brake is first applied by the SDS.”  Google therefore requests that NHTSA interpret S5.3 as met by this described approach.

We agree that the language of S5.3 requiring the service brake to be depressed does not necessarily require the service brake itself to be pressed or applied by any particular object or function, such as a human foot.  We also agree that if the SDS is controlling the service brake, in theory, it would be able to make the decisions that would accomplish the intent of this provision.  In order for NHTSA to assess compliance of Google’s vehicle with this standard, however, we would need more information regarding how the SDS applies the service brake.  We would also intend to develop and adopt through rulemaking performance standards and test procedures for evaluating how (for example) the SDS “determines the appropriate transmission position” and avoids “selecting a position other than park unless the service brake is first applied.”  NHTSA might also consider as part of such a rulemaking the safety intent of the standard, and how human occupants should be protected when the vehicle is making decisions about when to initiate movement.  Google may wish to petition the agency for an exemption from these provisions.[8] 

FMVSS No. 126, “Electronic Stability Control Systems”

FMVSS No. 126 contains performance and equipment requirements for electronic stability control (ESC) systems.  Google argues that because its vehicle will not have a steering wheel, and the SDS will control all aspects of steering, NHTSA should interpret the relevant provisions of FMVSS No. 126 “to allow compliance with the performance requirements of the standard to be tested on the basis of appropriate steering inputs provided by the SDS.”  Google cites in particular the definition of an ESC system in S4 as referring to “a means to monitor driver steering inputs” and the test conditions in S6 and test procedures in S7, which refer to steering wheel velocity and angle.

Test procedure provisions in S7 (such as S7.5.2, S7.5.3, S7.6, S7.6.1, S7.9.3, and others) refer specifically to measurement of “steering wheel angle,” which is, literally, a measurement of the angle of the steering wheel in degrees.  Without a steering wheel, we agree that Google cannot certify its vehicle design to such provisions.  As above, in order for NHTSA to assess compliance of Google’s vehicle with this standard, we would need to determine how to evaluate the SDS’ control of the steering inputs, and whether and how to modify test conditions and procedures to address more clearly the situation of a vehicle with steering controlled entirely by an AI driver, with no mechanism for the vehicle occupants to affect the steering.  Google could petition for exemption from the relevant provisions of FMVSS No. 126

and the agency could work to develop alternate test procedures.  Over the longer term, NHTSA would need to undertake rulemaking to incorporate test procedures into FMVSS No. 126 to provide a clearer path to compliance for similar future vehicle designs.

Google’s letter concludes with two attachments, summarily listing numerous additional provisions for which it seeks interpretation.  Attachment A lists “Requirements for which the ‘Driver’ or ‘Operator’ should be considered to be the Self-Driving System.”  The letter provides no further explanation or justification for those requested interpretations.  Attachment B provides a similar list of FMVSS “Requirements for which the ‘Driver’ should be considered to be a person seated in the left front designated seating position.”  We address these numerous items in Attachments A and B in Tables A and B, to this letter, respectively (attached).  The Tables in those attachments reproduce each of Google’s requests in the first three columns, and provide NHTSA’s response in the final column.

In closing, we note that, in some instances, it may be possible for Google to provide more information and explanations that would allow NHTSA to expand or otherwise revise interpretations set forth in this letter.  If Google believes it can address concerns and limitations expressed in this letter that prevent us from providing the full interpretation it seeks, we encourage the company to submit more information to do so. In addition, as discussed above, Google may wish to seek exemptions from standards and requirements addressed in this letter. I hope this information provided in this letter and its attachments is helpful.  If you have further questions, please feel free to contact me at (202) 366-2830, or Steve Wood, Assistant Chief Counsel for Vehicle Safety Standards and Harmonization, at (202) 366-2992.

                                                                        Sincerely yours,

                                                                        Paul A. Hemmersbaugh

                                                                        Chief Counsel

III.  Table A

NHTSA interprets the SDS to be the “driver” or “operator” for the following specific FMVSS provisions.  As discussed above, we have no defined way at this time of verifying Google’s compliance with this interpretation of these requirements.  Thus, if Google certified its compliance with these provisions consistent with this interpretation, NHTSA would be unable to conduct confirmatory testing to satisfy ourselves that such compliance was valid.  Therefore, unless and until NHTSA develops performance criteria and test procedures for evaluating whether the SDS satisfies the FMVSS provision in question to be met, or a standard providing equal or greater safety, it cannot interpret Google’s SDV as compliant with these standards and requirements.  Google may wish to petition the agency for exemption from these petitions in the interim.

1. Table B

For specific FMVSS provisions set forth on Google’s Attachment B, Google requested that NHTSA interpret the human occupant seated in the left front designated seating position (‘DSP’) as the “driver.”  As discussed above, NHTSA defines “driver” for purposes of the FMVSS at 49 CFR 571.3.  We also discuss above the need to amend that definition in light of the possibility that Google raises, that an AI driver could be the sole means of driving a vehicle.  Given the focus of the definition of “driver” as, essentially, the entity that controls steering, and given that Google’s proposed vehicle design gives the human occupant no means to steer the vehicle, the human occupant of the left front DSP could not be the driver.  Moreover, interpreting “driver” as the human occupant of the left front DSP in the instances below that Google cites would not, in most cases, provide Google with a means of certifying that its proposed vehicle design complies with the applicable standard, because the agency would need to establish standards and tests to verify that the design complies. 

It may be that for some of these provisions, the elimination of a human driver makes the provision unnecessary, as Google implies in its letter.  NHTSA will consider these issues further and looks forward to receiving additional information from Google to assist us in this consideration.  For the following provisions, we are therefore continuing to interpret “driver” as referring to the SDS, and discuss what steps would need to be taken next in order to provide a path forward:

[1] Google also submitted a supplemental letter date January 11, 2016, providing more detailed information on its approach to canceling the turn signal, which is discussed below.

[2] NHTSA considers this AI driver to be an item of motor vehicle equipment within the meaning of 49 U.S.C. 30102 and other applicable law and regulations.

[3] Google may also wish to reconsider its view that a pedal may never be needed in any circumstance, and that there is not a risk of harm associated with a pedal’s absence.

[4] Google may also wish to reconsider its view that the controls listed in Tables 1 and 2 of S5.1.1 may never be needed in any circumstance, and that there is not a risk of harm associated with their absence.

[5] We note that S9.4.1 of FMVSS No. 108 provides a mechanism by which a manufacturer can certify a “semi-automatic headlamp beam switching device,” but Google did not ask for interpretation of this provision, nor do we have enough information to assess whether Google’s proposed design would comply with this provision.

[6] Google may also wish to reconsider its view that a steering wheel and the vehicle occupants’ ability to control any lighting and signals may never be needed in any circumstance, and that there is not a risk of harm associated with their absence.

[7] Google may also wish to reconsider its view that rear visibility devices and systems may never be needed in any circumstance, and that there is not a risk of harm associated with their absence.

[8] Google may also wish to reconsider its view that a pedal may never be needed in any circumstance, and that there is not a risk of harm associated with a pedal’s absence.

[9] We note that NHTSA has previously interpreted this provision as prohibiting automatic (i.e., non-driver initiated) control of turn signal lamp flashing, but we do not consider this interpretation as inconsistent with those, because in this instance we are interpreting the “driver” (the SDS, given that the human occupants have no mechanism by which they can drive the vehicle) as being the entity controlling the vehicular hazard warning signal operating unit.  NHTSA would like to explore further with Google the instances in which Google would intend for the SDS to flash the hazard warning signals.

ID: 11673ZTV

Mr. James Baker
Technical Services Bureau
State of New York
Department of Motor Vehicles
Division of Vehicle Safety Services
Empire State Plaza
Albany, NY 12228

Dear Mr. Baker:

This is in reply to your FAX of March 14, 1996, to Taylor Vinson of this Office. You have assumed that vehicles "that bore a federal certification statement had SAE/Dot approved lighting", and believe that you are "finding out that may not be true." Specifically, you asked whether the agency "would accept lighting that does not bear the SAE/Dot lens markings. Eg. SAE STL-79, SAE DP-81."

The answer is yes, except for original and replacement headlamps covered by paragraph S7 of Federal Motor Vehicle Safety Standard No. 108 Lamps, Reflective Devices, and Associated Equipment ). The lenses of these headlamps must be marked DOT as a certification of compliance (paragraph S7.2(a)). These are headlamps manufactured for use on vehicles other than motorcycles.

The SAE codes you quote appear to relate to motorcycle lighting equipment other than headlamps. Thus, it may be that this type of motor vehicle is of some concern to you. Standard No. 108 contains alternative specifications for motorcycle headlighting systems. If the motorcycle has a headlamp system conforming to SAE J584 April 1964, no lens marking is required. However, if the motorcycle is equipped with one half of a type of headlighting system specified in S7, as it is permitted to do pursuant to paragraph S5.1.1.23 of Standard No. 108, the marking requirements of S7.2 continue to apply.

No DOT marking is required for the lens of any other original or replacement lamp. If a manufacturer wishes to certify compliance of a replacement lamp, paragraph S5.8.10 of Standard No. 108 permits it do so by putting a DOT symbol on the lens (otherwise the manufacturer is expected to certify by a label attached to the lamp, or a statement on the container in which it is

shipped). We have never required that lenses be marked with SAE code functions, and note that even the SAE materials that cover individual types of lighting equipment do not contain a specification that lighting devices be marked in accordance with SAE Recommended Practice J759 Lighting Identification Code though many manufacturers do so.

If you have any further questions you may refer them to Taylor Vinson of this Office (202-366-5263).

Sincerely,

Samuel J. Dubbin Chief Counsel ref:108 d:4/4/96

ID: aiam4515

Dear Ms. Black: Thank you for your letter of April l4, l988, providin further information about the Jaguar headlamp levelling system discussed in your letters of June and October l987 to which I responded on February 1, 1988. We support your efforts to call the driver's attention to the fact that the system does not automatically return to the 'zero' position from either of the two adjustment positions when those loading conditions no longer exist. Sincerely, Erika Z. Jones Chief Counsel;

ID: nht94-7.22

DATE: March 25, 1994

FROM: Roger W. Bruett -- Chief; Signed by Carol I. Morton -- Administrative Assistant, Equipment and Standards Review Unit, Washington State Patrol

TO: Chief Council, NHTSA

TITLE: None

ATTACHMT: Attached to letter dated 4/14/94 from John Womack to Carol I. Morton (A42; Std. 108)

TEXT:

It has come to our attention, new headlamps are being installed on BMWs, 750 models. These lamps are called XENON and are "Hella SSB 352 (lower beam) and Hella SSB 328 (upper beam) with replaceable bulb headlamp with glass lens and plastic reflector.

This office received a complaint from a BMW Dealership indicating an owner of one of these vehicles is being continually stopped by law enforcement because of these headlamps. The reason for concern from the officers, is the intensity of the light beam from these lamps is much whiter and appears to be stronger.

We would like to know if these headlamps are legal for use on motor vehicles. A report was sent to us stating they meet all the requirements of the FMVSS 108. This report was generated by ETL Testing Laboratory in New York.

I would appreciate your response on this subject as soon as possible.

ID: nht75-2.39

DATE: 05/30/75

FROM: AUTHOR UNAVAILABLE; E. T. Driver; NHTSA

TO: Stanley Electric Co., Ltd.

TITLE: FMVSS INTERPRETATION

TEXT: This responds to your letter of May 15, 1975, regarding Federal Motor Vehicle Safety Standard (FMVSS) No. 108 requirements for Type 1A and 2A automotive headlamps.

The following answers are provided for your specific questions

1. FMVSS No. 108 would not prohibit use of metal-back Types 1A and 2A headlamps that conform to all requirements of the standard.

2. The National Highway Traffic Safety Administration does not issue approvals on automotive equipment. The equipment manufacturer self certifies that the equipment conforms to the applicable FMVSS. The various states may, however, require equipment approval. Information on these approvals may be obtained from the American Association of Motor Vehicle Administrators, 1201 Connecticut Avenue, Washington, D. C. 20036.

3. The drawings of the Type 1A and 2A headlamp submitted with your letter, indicate that aiming pads have been deleted. Aiming pads are required by FMVSS No. 108.

For your information enclosed is a copy of FMVSS No. 108, which includes requirements for Types 1A and 2A headlamps, and a copy of Docket No. 75-8 Notice 1, that proposes to allow use of the four-lamp rectangular systems indefinitely.

ID: nht94-1.98

TYPE: Interpretation-NHTSA

DATE: March 25, 1994

FROM: Roger W. Bruett -- Chief; Signed by Carol I. Morton -- Administrative Assistant, Equipment and Standards Review Unit, Washington State Patrol

TO: Chief Council, NHTSA

TITLE: None

ATTACHMT: Attached to letter dated 4/14/94 from John Womack to Carol I. Morton (A42; Std. 108)

TEXT:

It has come to our attention, new headlamps are being installed on BMWs, 750 models. These lamps are called XENON and are "Hella SSB 352 (lower beam) and Hella SSB 328 (upper beam) with replaceable bulb headlamp with glass lens and plastic reflector.

This office received a complaint from a BMW Dealership indicating an owner of one of these vehicles is being continually stopped by law enforcement because of these headlamps. The reason for concern from the officers, is the intensity of the light beam from these lamps is much whiter and appears to be stronger.

We would like to know if these headlamps are legal for use on motor vehicles. A report was sent to us stating they meet all the requirements of the FMVSS 108. This report was generated by ETL Testing Laboratory in New York.

I would appreciate your response on this subject as soon as possible.