Industry Comments to NHTSA’s Federal Automated Vehicles Policy

In response to NHTSA’s Request for Comment on the Federal Automated Vehicles Policy (“FAVP”) published in the Federal Register on September 23, 2016, over 1,100 comments were submitted by industry participants, state and municipal transit agencies, policy groups, advocacy organizations and concerned citizens. These comments can be viewed at www.regulations.gov on the docket NHTSA-2016–0090.

This article highlights key points made by the auto and technology companies primarily involved in developing highly automated vehicles (“HAVs”), and does not purport to summarize or include all comments submitted. Comments are discussed in three categories: (1) Vehicle Performance Guidance; (2) Model State Policy; (3) Regulatory Tools. All of the industry participants commended NHTSA on its efforts to promote self-driving vehicles and to accelerate development, and noted the tremendous safety benefits of HAVs. This article focuses on the suggestions and comments offered to the policy and does not repeat these points.

Section I — Vehicle Performance Guidance

Self-Driving Coalition For Safer Streets. The Self-Driving Coalition for Safer Streets (“Coalition”) was established by Ford, Google, Lyft, Uber and Volvo in April 2016. The Coalition believes the proposed Vehicle Performance Guidance has three potential issues: (1) manufacturers may need to submit more than just an initial Safety Assessment letter leading to “repeated re-submissions outside of significant updates to HAVs”; (2) these Safety Assessments may lead to disclosure of proprietary confidential information; and (3) the Guidance does not “affirmatively discourage states from mandating the use of this guidance,” which is voluntary at the federal level.

Specifically, the Coalition requests that confidential business information not be collected and that updates required by the Safety Assessment process be limited. The Coalition contends that updates to the Safety Assessment letter should not be required unless a “significant update” is made, which should be defined by NHTSA to mean “a substantial change to the HAV system that affects its capability or safety (e.g. a wholly re-defined ODD or a step level increases in the system’s level of automation) with a company making the determination whether the change is indeed substantial.”

With respect to the Safety Assessment letter, the Coalition proposes that “flexibility be a hallmark of the Safety Assessment letter process” in order to allow manufacturers to choose how to submit letters across features or systems in common or across multiple vehicle models. Flexibility would also allow the manufacturer to “minimize the number of letter submissions” during the testing phase, in light of numerous software updates that would likely occur. Finally, the Coalition suggests flexibility would allow for differences in submissions for Level 2 versus Level 4 vehicles, which have different requirements.

Google. Google’s suggestions regarding the Guidance fall into three categories: (1) streamlining the process for submission and review of Safety Assessments; (2) clarifying terminology and definitions; and (3) specific safety assessment areas.

First, Google recommends that NHTSA’s review of Safety Assessments “should not result in substantive or technical judgments on HAV system design or other product decisions” and the review should only confirm that a submitter’s Safety Assessment “has explained how its processes and practices reasonably address each of the 15 areas.” Regarding process, Google recommends that NHTSA clarify that its review of Safety Assessments “(1) will be completed in a maximum of 30 days; (2) will thereafter immediately result in a response letter to the submitter stating either that (a) the submission is complete, or (b) the agency would like further information on specific Safety Assessment items; and (3) if the submitter provides additional information, it will be treated and reviewed in the same manner as the initial submission and a response letter will be provided by the agency within 15 days.” Google also makes several suggestions regarding public disclosure and confidential business information. Google suggests that NHTSA clarify “whether it expects submitters to provide confidential business information” in its Safety Assessment, and if so to require that any confidential business information submitted be attached as “an appendix separate from the Safety Assessment letter itself.” Google requests that the Safety Assessment letters (without the appendices) and NHTSA’s responses be published simultaneously without revealing any confidential information. Google also asks NHTSA to clarify that it “intends to treat all confidential commercial and financial information submitted in connection with Safety Assessments” as “voluntarily submitted” under Part 512. Regarding submission of Safety Assessments, Google recommends that NHTSA clarify that “only major changes to the HAV system that affect its capability or safety (e.g., a wholly re-defined ODD or a step-level increase in the system’s level of automation) warrant submission of an updated Safety Assessment.” Google also urges NHTSA to clarify which entities involved in the development, manufacture and implementation of HAV systems would need to file a Safety Assessment. Google recommends the following entities be required to make submissions: (1) the motor vehicle manufacturer where the HAV system is designed for and installed in a new vehicle prior to first sale (not resale); (2) the HAV system manufacturer where the system is added to a vehicle after its first sale (not resale); and (3) the entity which is not a HAV system or vehicle manufacturer but which operates HAVs in a commercial or other service, which should make a submission as to the relevant Safety Assessment items that pertain to their role (e.g., data recording, privacy, consumer education, etc.).

Second, on terminology and definitions, Google urges NHTSA “to consider not using the term ‘operator,’ and instead using SAE J3016’s terminology to distinguish between roles of individuals related to driving automation” and if “operator” is used, then NHTSA should clarify “that for a vehicle with an SAE Level 4 or 5 system used as a part of a commercial service, the ‘operator’ is defined as the entity operating the service, and should not include occupants who are mere passengers.” Google also recommends that NHTSA clarify that “deployment” refers to “the use of HAVs by members of the public who own, lease, rent or pay a fee to ride in the HAV.”

Third, on specific Safety Assessment areas, Google supports the agency’s proposal to explore a mechanism to “facilitate anonymous data sharing” among parties testing/deploying HAVs and recommends a dialogue between the agency and the industry as to what types of data would be appropriate and useful to share, the format and compatibility of the data and mechanisms to protects proprietary and sensitive data. Google also recommends a change to the section on ethical considerations, taking issue with the current NHTSA guidance as suggesting that “some or all safety-critical software elements of an HAV system should be effectively crowdsourced.” Google suggests that NHTSA should instead require submitters “to indicate in their Safety Assessments that they have a process in place to make reasoned decisions on ethical considerations regarding HAV design that take into account the interests of relevant road users, including drivers, passengers, and other anticipated road users (e.g., pedestrians, bicyclists, etc) within the vehicle’s ODD.”

Ford Motor Company. Ford submitted a proposed Sample Assessment Letter and provided specific comments on the information requested and format of the Safety Assessment letter process. In particular, Ford recommends that NHTSA create a dedicated site for sharing of public information and consider ways to simplify the voluntary process of submitting Safety Assessment letters and handling confidential business information. Ford asks NHTSA to provide flexibility with the basis for submission and to limit the Safety Assessment letter submission requirement to the “testing/deployment/revisions for an automated system” without “duplicating the efforts for new vehicle model introductions.” Ford believes “multiple submissions during the testing phase are not feasible” and suggests a phased approach to submissions.

Volvo Car Corporation. Volvo believes that by applying the Guidance to Level 2 that it applies immediately to such vehicles currently on the market. Volvo is concerned that “this retroactive application is impracticable” and that NHTSA should instead “use its existing authority for SAE level 1 to 2 vehicles and the federal guidelines should only apply to SAE levels 3 to 5.” Volvo also notes the guidance seems to apply to original equipment, replacement equipment and software updates, which could create delays for “necessary updates and replacements,” particularly in testing programs “because testing typically includes continual and frequent changes and updates.” Volvo objects to the four month waiting period for AV testing from the time an OEM completes the assessment to the time testing can begin, noting that this will delay “AD testing in the US while it will be happening in other countries.” Regarding data recording and sharing, Volvo supports the goal to “collect and share relevant statistical data that increases public awareness and trust in the safety of HAVs” but believes “it is critical that specific metrics be defined and more research be done in order to create a valuable definition of positive outcome events.”

Uber Technologies. Uber believes there are three “foundational principles” that should guide regulation in this area that are “overlooked by the current NHTSA guidance”: (1) self-driving software is “fundamentally different” from “traditional vehicle equipment software” because it is continuously updated; (2) self-driving technology will “facilitate a fundamentally different mode of vehicle operation and use” (ridesharing or commercial fleets) and (3) self-driving technology “does not warrant a fundamental re-balancing of powers between local and federal authorities” regulating vehicle safety.

First, regarding the regulation of self-driving software, Uber notes that “the proposed four-month review process may be sensible for embedded software that cannot be easily updated after a vehicle is sold, but for self-driving software that might be updated daily, it will have the perverse effect of hindering deployment of safety-related and other upgrades.” Uber suggests that instead of “establishing a burdensome oversight process, the agency’s focus should be on defining outcomes and testing for achievement of those outcomes — and allowing companies to innovate different ways of achieving them.”

Second, Uber expresses “disappointment” that there is “so little focus on fleet operators and fleet development of self-driving vehicles, even though that category holds the potential for faster and safer deployment of self-driving technologies.” Uber states that “self-driving technology has the potential to help us realize a future in which car ownership is the exception, not the rule.” Uber notes that the “traditional automakers have also indicated they are interested in pursuing various types of fleet models” but that NHTSA’s guidance “does not reflect the paradigm shift that self-driving vehicles are poised to enable from car ownership to car use.” Uber states that the distinction between “testing” and “deployment” in the guidance only envisions a sale-to-consumer model and “falls short of an appropriate regulatory model for self-driving fleets.” Uber also suggests that fleets may have an “operator” in the car trained on the vehicle’s technology which would be “a critical distinction from vehicles controlled by ordinary consumers, who, as NHTSA has recognized, are prone to the safety risks that come with driver complacency, distraction, impairment, and operational misuse. Put simply: fleet managed, operator-assisted vehicles do not warrant inclusion in NHTSA’s guidance, regardless of whether they are equipped with self-driving technologies, or not.”

Third, Uber contends that “regulating drivers who operate vehicles has always been the purview of individual states, and outside the federal authority of NHTSA.” Uber states that NHTSA should not take away this power from states, and that states should be allowed to continue “to investigate and pilot different approaches as they deem appropriate.” Uber notes that given how quickly the technology is evolving, “it is premature to try to mandate a nationalized framework and shift responsibilities away from the states.” Uber suggests that NHTSA’s efforts should be focused elsewhere and that “elaborate guidance or hasty regulation of self-driving vehicles does not make sense right now.” Uber requests “a thorough cost-benefit analysis” to help the agency prioritize its regulatory efforts and believes such an analysis will show that “this guidance will stifle innovation without a corresponding safety benefit.”

Lyft. Lyft is “concerned that the guidance does not set forward a clear statement that NHTSA is neither seeking, nor interested in seeking” consumer use data, including personally identifiable information, usage patterns, geo-location data, and other aspects of ride data. Lyft urges NHTSA to make clear it is not gathering specific consumer data collected by any ridesharing platform, in order to avoid any appearance that “big government” is sifting through personally identifiable information, which would erode public trust. Lyft suggests the guidance should formally recognize ridesharing and alternative industry models, which will “democratize the benefits” of HAV technology and which may be able to address issues such as consumer education in innovative ways.

Apple. Regarding the Safety Assessment process, Apple notes that as written the Safety Assessment process “could be interpreted as requiring pre-approval by NHTSA prior to testing” and if NHTSA does not intend a pre-approval requirement that NHTSA should clarify this provision. In addition, Apple notes that rapid iterations within the four month period are likely and should not require re-submission of multiple Safety Assessments. Apple suggests the process for Safety Assessment should be different for “internal development vehicles” than for deployment vehicles, and should consist of the following steps: (a) discuss “a test plan and acceptance criteria that may include multiple levels of functionality” with NHTSA before submitting Safety Assessment; (b) document the agreed-upon test plan and acceptance criteria, including the planned timing for public road testing, in the Safety Assessment; and (c) notify NHTSA when public road testing starts or if the test plan changes.

Regarding data collection and sharing, Apple “agrees that companies should share de-identified scenario and dynamics data from crashes and near misses” and that the data should be “sufficient to reconstruct the event, including time series of vehicle kinematics and characteristics of the roadway and objects.” Apple believes this data sharing will allow the industry to “build a more comprehensive dataset than any one company could create alone.” Apple states that “data sharing should not come at the cost of privacy” and believes that “companies should invest the resources necessary to protect individuals’ fundamental right to privacy.” Apple “urges NHTSA, industry partners and relevant federal agencies to continue to address privacy challenges associated with the collection, use and sharing of automated vehicle data.”

Apple “commends NHTSA for including ethical considerations in the Policy” and notes the “broad and deep human impact” of automated vehicles. Apple “strongly affirms the need for thoughtful exploration” of these ethical issues. Apple notes three areas that should be considered: (1) “the implications of algorithmic decisions for the safety, mobility and legality of automated vehicles and their occupants”; (2) the challenge of “ensuring privacy and security in the design of automated vehicles”; and (3) the “impact of automated vehicles on the public good, including their consequences for employment and public spaces.”

General Motors. GM suggests several changes to the safety assessment process. GM believes the Safety Assessment submission should be different for various stages of testing and deployment, with some of the 15 guidance areas applicable in testing and others not. GM also advocates refining the acknowledgements from the“meets, does not meet, not applicable” framework to “a general discussion of relevant portions of the guidance topics,” since NHTSA does not intend the Safety Assessment to be utilized as a certification or pass/fail, but rather a “flexible way for submitters to provide assurance to NHTSA and to the public that the vehicles are safe.” GM notes that the “meets/does not meet” framework could be confused with compliance testing, and since there are no objective criteria for these tests, they “cannot be assessed in pass/fail terms. GM believes Level 2 vehicles should not be included in the guidance, only Levels 3–5.

Tesla Motors. Tesla believes NHTSA should “remove the four-month delay between submission of a Safety Assessment and commencement of road testing” as such a delay will not produce “any measurable benefit” and will “unnecessarily slow the development process of HAVs.” In addition, this delay would reward manufacturers who do not comply with the voluntary Safety Assessment process.

Mercedes-Benz USA. Mercedes notes that the “new voluntary requirements apply to both testing and deployment” which is “unprecedented” since OEM’s have been allowed to test non-compliant vehicles throughout the U.S. without obtaining prior permission from NHTSA.

Mercedes also objects to the “retroactive application of this guidance to products we have had in the US market since model year 2014, and which have performed satisfactorily in customers’ hands.” Mercedes states that covering Level 2 functionality would require it “to assess retroactively whether our existing and widely-deployed vehicles ‘meet’ this new guidance in all particulars” and the mere fact of documenting “that an existing product ‘does not meet’ NHTSA’s voluntary guidance — regardless of inconsequentiality — is sufficient to create unsustainable business and liability risk for manufacturers, even if the products have consistently performed safely in the field.”

Regarding data collection and sharing, Mercedes notes that the FAVP “loosely specifies the gathering of massive quantities of data on both crashes and ‘near crashes’” and it is “unclear which data NHTSA intends to be collected.” Mercedes contends that such a data collection requirement is “neither practicable (the collection and storage requirements are cost-prohibitive and exceed the capacity for on-board storage or wireless transfer) nor reasonable, as unrestrained sharing of competitive information and liability-sensitive data with authorities and the general public would violate antitrust, privacy and security laws and policies established for the protection of the public interest.”

Mercedes “strongly recommends that NHTSA remove the ‘ethical considerations’ item from the 15 point list of topics that must be addressed in a safety assessment letter” although it supports and is “eager to contribute to continued dialogue” to explore the development of “a reasoned approach to the ethical questions related to deployment of HAVs.”

Mercedes notes that the “meets/does not meet” framework is unworkable due to the “lack of objective criteria provided by the descriptions of the 15 areas of safety concern.” Mercedes also objects to the process of re-submitting safety assessments for most product changes and the four month delay every time such resubmission is required. Mercedes believes “only updates that change the manner in which vehicle motion control is accomplished (e.g., a new sensing technology is added or a control technology is changed) should require re-application.”

BMW. BMW suggests the Guidance should only apply to vehicles being deployed and not vehicles being tested, as FMVSS compliance is not currently required for test vehicles. BMW also objects to the retroactive application of the Guidance to Level 2 vehicles and believes it should not apply to systems already deployed in the market. BMW urges NHTSA to dissuade states from making the Guidance mandatory by “unilaterally issuing their own AV regulations based on elements of the FAVP.” BMW suggests that submitting Safety Assessment Letters for test vehicles “would increase the burden with little to no benefit.” BMW believes NHTSA should work with the industry to create a template for the Safety Assessment for each of the various SAE levels and that only “changes to the Operation Design Domain or the SAE level would qualify as a material change and therefore warrant a new Safety Assessment Letter submission.” BMW believes the nature of the layout of Safety Assessments is not suited to public release and “may have the possibility of mischaracterizing the ADS in question as being incomplete or not suitable for road use” and suggests instead an executive summary format that could be released publicly.

On data recording and sharing, BMW notes that events the ADS avoided are not recorded and that data should be limited to a “crash or other physical occurrence that meets or exceeds a trigger threshold” because defining a “near miss” is subjective. BMW suggests sharing should be “limited to production vehicles and corresponding events” not events during the testing phase.

Regarding ethics, BMW recommends that the “development of algorithms for defining ADS behavior not be made public” as they will likely “contain proprietary and confidential business information. BMW does, however, see the potential for decision rules to be described generally to the public.”

Toyota Motor Company. Toyota notes that the 15 areas for safety assessment are not stated in objective terms and therefore “it is not appropriate to attempt to state whether they have been ‘met’, ‘not met’ or are ‘not applicable’.” Toyota also believes that varying levels of detail should apply in submissions to protect confidential information, and that responses will vary between testing and deployment phases. Toyota objects to the frequency of submissions based on the “significant update” standard, as it will depend on one’s “understanding of what is ‘significant’” and manufacturers will need to “use their best judgment in determining what changes” are significant. Toyota suggests that the four month waiting period should be removed and that companies should be able to begin/resume testing as soon as a letter or update has been submitted.

Regarding data collection and sharing, Toyota believes the requirements should be different for testing and deployment. During developmental testing, Toyota notes that data is likely to contain “proprietary information, methods and practices” and may contain components from third parties where sharing would violate agreements. Toyota notes that “testing is a highly iterative process, and its purpose is to essentially ‘break’ the system and find as many ‘corner cases’ or ‘edge cases’ as possible to improve the system.” Therefore “reports of disengagements or even near-crash activities during testing would be detrimental to customer acceptance if made public, because it is not representative of the capabilities of the system ultimately put into production.”

Honda Motor Company. Honda believes the Guidance should apply only to Levels 3–5, as it is not clear how the components would apply to Level 2 vehicles. Honda also notes retroactive application to Level 2 systems already sold in the market would require significant resources. Honda notes that application of the safety assessment process to test vehicles is “potentially problematic” because it is not possible provide assurance that all aspects of the assessment have been satisfied when a vehicle is by definition “still under development.” Moreover, Honda notes that having to resubmit the assessment every time an improvement is made during development would be burdensome. Honda also notes it is “unreasonable to expect immediate conformance with the policy guidance” without any lead time for adoption and implementation. As to the ethical requirement, Honda notes that it should not apply to Level 2 vehicles, and notes that NHTSA should work with the industry through SAE or CAMP to develop guidance for manufacturers. Honda notes that “it will create confusion for all road users unless there is some unified consensus around typically acceptable behavior in a variety of reasonably foreseeable circumstances.”

Zoox. Zoox notes that “a testing paradigm is very different from a deployment paradigm” and that NHTSA should develop different assessment templates for testing and deployment, as well as for different HAV classes. Regarding data collection and sharing, Zoox believes NHTSA should specify “the kind and scope of positive outcome data it expects to see” and that sharing of data should depend on the type of data. Zoox can “see value in sharing corner case data because it can improve safety across the industry” and thinks the industry “should develop a framework that enhances safety by perhaps establishing a neutral clearinghouse to collect, anonymize and standardize the data.” Regarding ethical considerations, Zoox believes “it is premature to develop a full suite of standards around ethical vehicle behaviors, other than to embed core safety-first principles into development.” Zoox states that “our view is that existing legal frameworks exist to disambiguate ethical considerations in incidents after they occur.”

Section II — Model State Policy

Self-Driving Coalition For Safer Streets. The Coalition expresses concern that the Model State Policy issued by NHTSA risks “encouraging states to embark upon (or carry out plans for) overly-restrictive and incongruous legislative action in this area.” The Coalition calls on NHTSA to “use this opportunity to affirmatively discourage states from embarking on creating a patchwork of inconsistent laws and regulations that will stifle this emerging industry” and to “leverage its existing position as the federal vehicle safety authority to safeguard against overlapping regulation by state and local governments.”

Google. Google also recommends that NHTSA revise the policy “to remove language which could be interpreted to invite States directly to regulate HAV systems and vehicles, or otherwise encourage inconsistent regulatory approaches among States, or between States and the federal government.” Google asks NHTSA to state “unequivocally” its intention that States “refrain from regulating in the area of HAV safety.” As to the specific Model State Policy language concerning HAV testing, Google asks NHTSA to clarify that “HAV systems may be tested without a person physically in or on the vehicle, as long as the testing entity retains responsibility over the vehicle and the vehicle can achieve a minimal risk condition if necessary.” Google also recommends that NHTSA clarify that “an application for testing need only state that the vehicles used for testing have been certified by the manufacturer as meeting the application FMVSS, and that NHTSA does not intend through its Model State Policy for States to extend FMVSS compliance obligations beyond those parties that are regulated under federal law.” Finally, Google urges NHTSA to revise the Model State Policy to “recommend to States that they grant appropriate recognition to HAV testing and deployment that has occurred in other States.”

Ford Motor Company. Ford is concerned that “absent clear encouragement otherwise, States will move to make the voluntary assessment process mandatory.”

Volvo Car Corporation. Volvo notes that the NHTSA guidance “does not prevent a patchwork of state policies on testing and operation” and that states are “continuing to act.” Volvo recommends that NHTSA “discourage state activity in order to prevent a patchwork and allow OEMs to self-certify to the NHTSA guidance rather than a separate state by state approval process.”

Lyft. Lyft notes that states will likely be “active in regulating HAVs irrespective of any request from NHTSA to refrain from such action” and therefore suggests the model state policy “should directly encourage States to adopt reciprocity provisions that would accelerate the ability for deployment of HAVs across the nation for vehicles whose operational design domains have already been safely and successfully tested or deployed elsewhere.” Lyft also suggests removing from the model state policy “the need for HAV operators to obtain a special license” which adds “unnecessary burdens to the process” and “imposes potential impediments to those with disabilities.”

Apple. Apple believes the federal government should “maintain sole authority over the safety of motor vehicles” and that states should adopt the model policy to “avoid policy proliferation and inconsistencies that may prevent or delay deployment.”

General Motors. GM notes that the FAVP suggests that states use a pre-approval approach. GM disagrees with “state-level pre-approval of vehicle safety performance because this will invite conflicts between NHTSA’s role and the states’ roles.”

Tesla Motors. Tesla supports NHTSA’s efforts “to provide smart federal policy for HAVs.” While Tesla would have preferred an “innovation before regulation” approach, it now believes “the current threat of patchwork state-by-state regulation justifies NHTSA’s intervention related to a technology that has yet to be deployed.” Tesla applauds “NHTSA for asserting its leadership in the face of states issuing well-intentioned, but misguided, legislation aimed at HAVs.” Tesla believes that state-level initiatives create uncertainty for developers, blur the traditional division of regulatory responsibility, and create “barriers to the testing and deployment of HAVs where no such barriers previously existed.”

Tesla notes that the FAVP as drafted “could reasonably be read” as encouraging states to create new laws and regulation rather than discouraging them. Tesla urges NHTSA to “revise the Policy to explicitly provide that states have no reason to create regulations related to HAVs, and thus should avoid doing so.” In addition, Tesla believes “NHTSA must clarify whether the Policy is truly voluntary in nature” and should not “attempt to use the Model State Policy as a patchwork means of mandating NHTSA’s voluntary Vehicle Performance Guidance.” Tesla points to the revised draft regulations published by the California DMV which would mandate permits to test and deploy HAVs and “to obtain a permit, a manufacturer would have to certify compliance with the Vehicle Performance Guidance and submit a copy of NHTSA’s Safety Assessment.” Tesla also believes NHTSA should “more explicitly discourage states from enacting regulations that modify the Model State Policy, particularly respecting performance standards.” Tesla notes such performance standards should be preempted by new FMVSS that NHTSA has indicated it will issue.

Mercedes Benz USA. Mercedes notes that “the FAVP can be read as encouraging states to adopt this Guidance as mandatory regulations” and believes that “NHTSA should clarify that its guidelines are in fact voluntary guidelines, and that states should not impose them as de facto regulations.” Mercedes states that “far from discouraging the proliferation of varying state regulations of HAVs, the NHTSA guidance encourages states to ‘experiment’ in how they regulate the testing and deployment of HAVs” which will undermine NHTSA’s stated goal of “preventing the development of a patchwork of varying state regulations for HAVs.” Mercedes recommends that NHTSA clarify its potentially conflicting recommendations to states and that “it explicitly discourage states from imposing any vehicle certification requirements for HAVs, whether for testing or deployment.” Mercedes also notes that state distracted driver laws should be clarified so they do not apply to “the users of HAVs while a level 3–5 automated driving feature is engaged.”

BMW. BMW agrees with NHTSA “in strongly encouraging states to allow NHTSA to continue using its authority to regulate the performance of ADS-equipped vehicles” and for states to continue their role of “promoting roadway safety by regulating licensing, registration, traffic laws, safety inspection and insurance.”

Toyota Motor Company. Toyota believes NHTSA should “reiterate that States should not attempt to codify or enforce this Guidance as it related to performance of HAVs.”

Honda Motor Company. Honda encourages the DOT to “work with state governments to help ensure that the provisions of the FAVP are not prematurely made statutory by states inappropriately requiring compliance with the Vehicle Performance Guidance for Automated Vehicles.”

Zoox. Zoox believes states should “evaluate their current laws and regulations to address unnecessary impediments to the safe testing, deployment and operation of HAVs.” Zoox also recommends that states “work together to standardize and maintain road infrastructure” and also “ensure that updates to traffic laws are submitted in machine readable format to a secure database that could automatically push updates to AV developers.”

Faraday Future. Faraday suggests that the model state policy is unnecessary as several states are taking a wait and see approach or allowing testing without the suggested application process. Faraday notes the model policy may not promote uniformity as was intended and may instead result in disparities among states.

Section III — Regulatory Tools

Self-Driving Coalition For Safer Streets. The Coalition calls for two key actions beyond the FAVP to facilitate deployment of self-driving vehicles: (1) NHTSA should pursue narrow rulemakings to amend the FMVSS which currently mandate human operator controls; and (2) Congress should enact legislation to facilitate the rapid deployment of HAVs, including revising NHTSA’s exemption authority to allow a greater number of vehicles to be tested on public roads (currently capped at 2500 vehicles for up to two years).

The Coalition also opposes other actions proposed by NHTSA, including “any proposal to obtain pre-market approval authority for the deployment of HAVs.” The Coalition supports the continued use of the self-certification approach “that has been the hallmark and foundation of automobile regulation in this country for the past several decades.”

Google. Google strongly supports legislation “expanding NHTSA’s exemption authority to eliminate the temporal (two year) and production (2,500 vehicles annually) limits currently imposed by statute.” However, Google has “significant concerns about proposals to introduce pre-market approval and hybrid certification/approval (collectively, pre-market/hybrid approval)” and agrees with the NHTSA language noting such a system “would be a wholesale structural change in the way NHTSA regulates motor vehicle safety and would require both fundamental statutory changes and a large increase in Agency resources.” Google is concerned that such changes would delay the introduction of “life-saving technologies” and would be inappropriate in light of the lack of “existing standards and testing protocols,” which would lead to a reliance on “engineering judgment” rather than “established performance metrics, thresholds, and test procedures and equipment specified in regulations.” Therefore, Google recommends that NHTSA “actively oppose any legislation seeking to establish a pre-market/hybrid approval regime in the United States.”

Ford Motor Company. Ford does not believe “a type-approval or pre-market certification model will improve safety” and believes that the self-certification framework “can be successfully adapted to suit mass deployment of AVs.” Ford supports an increase in the exemption limit to allow for the “deployment of automated vehicles at a suitable scale.”

Volvo Car Corporation. Volvo recommends against moving to pre-market approval, which would be a “dramatic shift away from self-certification which has been the basis for US auto regulation.” Volvo believes pre-market approval “would delay safety advancements and could create public cynicism regarding HAV technology.” Volvo also believes “it would be impracticable for the agency to regulate post-sale software updates” and requiring approval would cause delays in rolling out over the air safety updates.

Lyft. Lyft supports expanding NHTSA’s exemption authority “to facilitate the safe testing and introduction of innovative new safety features in newly redesigned HAVs.”

Apple. Apple notes the FAVP suggests that “companies must seek an exemption for public road testing of internal development vehicles not fully compliant with FMVSS.” Apple states that the FAST Act allows established manufacturers, but not new market entrants, to test on public roads without seeking exemptions from FMVSS, and requests that new entrants to the market be treated the same as existing manufacturers. Apple requests that NHTSA amend the FAVP “to state that exemptions are not required for the controlled testing of internal development vehicles on public roads, provided they will never be used by the general public.”

General Motors. GM notes that guidance to the states should reflect that vehicles should meet applicable federal laws, not just FMVSS, as other laws such as the FAST Act may apply. GM supports NHTSA’s proposed use of interpretations for HAVs and suggests a 90 day time limit should be firmly established. GM also encourages more timely action on Part 555 exemptions and rulemaking.

Tesla Motors. Tesla believes that “pre-market approval authority would neither promote consumer acceptance nor ensure the safety of HAVs better than NHTSA’s existing regulatory authority.” Instead, Tesla “firmly believes that pre-market approval authority would impede, rather than expedite, innovation.” Tesla notes that pre-market approval can be appropriate for “mature technologies” but is “inappropriate for use in evaluating the safety or compliance of emerging technologies.” Tesla points out that type approval has been used in Europe for technologies “where clear measures exist by which the type approval authority can assess safety.” In contrast, “for new technologies, such as those that are being introduced with HAVs, NHTSA does not have clear metrics against which to judge the safety of the technology.” Since each manufacturer “will likely choose to implement its technology in different ways, at least initially,” Tesla notes it would be impossible to create a program against which such systems could each “receive objective pre-market approval” and choosing a single set of criteria would stunt innovation. Additionally, requiring pre-market approval would slow HAV development.

Tesla supports NHTSA’s request to receive expanded exemption authority and believes it “will better support the development of HAVs and similar new technology.”

Mercedes-Benz USA. Mercedes notes that NHTSA can apply different rules to technology companies than to established car manufacturers and suggests that this distinction should apply “for test vehicle requirements generally.” Mercedes contends that “established vehicle manufacturers, who have demonstrated their ability to safety test new vehicle technology on public roads over many decades, should not be required to submit safety assessment letters and updates for their test vehicles equipped with level 2–5 driving automation features.”

Regarding additional regulatory tools, Mercedes “does not agree that NHTSA has or should request authority over product design and development decisions for which HAV manufacturers and their suppliers bear complete reputational and product performance responsibility (including legal liability).” Instead, Mercedes believes “NHTSA’s role should remain that of setting minimum, objective performance standards and test procedures for safety that leave product design and development to HAV manufacturers and their suppliers.”

Mercedes does not believe that granting NHTSA pre-market approvals for motor vehicles and equipment is “good public policy” and notes that self-certification is preferred and “has done well for 50+ years.” Mercedes notes that the comparison to FAA and oil industry type approval is incomplete without mentioning that those industries “also enjoy legal liability protection conferred by Congress.”

BMW. BMW welcomes discussion with NHTSA “regarding the possible benefits of an EU type approval system to promote harmonization and/or equivalency.”

Honda Motor Company. Honda believes that adoption of a type approval approach “could significantly impede the ability of a manufacturer to bring a product to market.” Honda believes any such pre-approval “could be highly subjective” and “opens the door for introduction of potential biases” since there is a lack of “clear, well-defined performance criteria.”

Zoox. Zoox believes it is premature to consider proposals concerning pre-market approval and that such a system “has the potential to stall industry-driven innovation.” Zoox supports expanding exemption authority for NHTSA.