Earlier today, the National Highway Traffic Safety Administration (NHTSA) released its long-awaited Federal Automated Vehicles Policy. An op-ed published in the Pittsburgh Post-Gazette under President Obama’s name touted the nonbinding guidance document spouting platitudes and throwing out the obligatory rah-rah-regulation trope that largely characterizes his administration:
There are always those who argue that government should stay out of free enterprise entirely, but I think most Americans would agree we still need rules to keep our air and water clean, and our food and medicine safe. That’s the general principle here. What’s more, the quickest way to slam the brakes on innovation is for the public to lose confidence in the safety of new technologies.
Actually, Mr. President (or rather, his ghostwriter), the quickest way to slam the brakes on innovation is for bumbling bureaucrats to outlaw it.
But by and large, NHTSA’s guidance document is a relatively measured approach to automated vehicle technology and makes some important positive contributions to the public policy environment.
The guidance is divided into four sections:
I. Vehicle Performance Guidance for Automated Vehicles, which are recommendations aimed at manufacturers and mostly provide for standardizing documentation;
II. Model State Policy, which discusses the separate roles of federal and state governments in regulating motor vehicles;
III. NHTSA’s Current Regulatory Tools, which discusses the mechanisms authorized under current statute that can speed testing and deployment of automated vehicles, especially letters of interpretation and exemptions; and
IV. Modern Regulatory Tools, which discusses a wish-list of potential regulatory mechanisms NHTSA would consider beneficial.
The good news for fans of automated vehicles is that NHTSA appears to recognize the harm that government can do in delaying or adding cost to the deployment of automated vehicles: keeping less-safe vehicles on the road longer than would otherwise be the case translates to increased traffic crashes, injuries, and deaths.
Another positive change was the agency discarding its levels of automation (1-4) published in its 2013 Preliminary Statement of Policy in favor of the SAE International’s levels of automation (0-5), summarized as:
- At SAE Level 0, the human driver does everything;
- At SAE Level 1, an automated system on the vehicle can sometimes assist the human driver conduct some parts of the driving task;
- At SAE Level 2, an automated system on the vehicle can actually conduct some parts of the driving task, while the human continues to monitor the driving environment and performs the rest of the driving task;
- At SAE Level 3, an automated system can both actually conduct some parts of the driving task and monitor the driving environment in some instances, but the human driver must be ready to take back control when the automated system requests;
- At SAE Level 4, an automated system can conduct the driving task and monitor the driving environment, and the human need not take back control, but the automated system can operate only in certain environments and under certain conditions; and
- At SAE Level 5, the automated system can perform all driving tasks, under all conditions that a human driver could perform them.
This shows the agency is aware that conflicting definitions and terminologies were creating unnecessary confusion among developers, insurers, regulators, and lawmakers. Most of the recommendations contained in the guidance document focus on highly automated vehicles (SAE Levels 3-5).
But there is room for improvement. Below are a number of highlights and lowlights with NHTSA’s current guidance sections and potential future problems that may develop. They are categorized under the four policy sections listed above.
I. Vehicle Performance Guidance for Automated Vehicles
This section is relatively straightforward and contains few new problematic policy positions. For example, it notes that entities previously not considered to be manufacturers, “e.g., alterers and modifiers,” under the National Highway Traffic and Motor Vehicle Safety Act of 1966’s Federal Motor Vehicle Safety Standards (FMVSS) should try to follow the policy statement in the same way as traditional manufacturers, with the agency recognizing that aftermarket changes to hardware and software could dramatically alter the vehicle. Part of this process involves NHTSA conducting a review of manufacturers’ stated SAE levels of automation.
Even better, the guidance document highlights an issue that has been raised numerous times over the past few years: the ability of automated vehicles to temporarily violate driving laws when needed. For example, if an accident or double-parked vehicle blocks a lane, should the automated vehicle stop and wait for the lane to be cleared, or should it cross a double-line lane marking in violation of the rules of the road? To a rational person, the answer is simple: drive around the obstruction if it is safe to do so. The law permits these temporary violations. Fortunately, NHTSA endorses such a policy.
The trouble with the vehicle performance guidance begins with the safety assessment letter NHTSA asks to be submitted to the agency’s Office of Chief Counsel. The letter would detail all elements to be included in a safety assessment, including data recording, cybersecurity, operational design domain, object and event detection and response, and the minimal risk condition (fall back to a safe state in the event of a system failure). NHTSA repeatedly maintains that such compliance and reporting is entirely voluntary—as is the case given this is a nonbinding guidance document.
However, in the following section, Model State Policy, NHTSA recommends that states require their automated vehicle registration agencies to include in the “[registration] application…that each vehicle used for testing by the manufacturers or other entities follows the Performance Guidance set forth by NHTSA…” (p. 41)
NHTSA cannot credibly say it is merely recommending voluntary, nonbinding actions and then turn around and tell other government agencies to mandate them. If NHTSA wishes to mandate automated vehicle performance safety assessments, it should go through the normal rulemaking process as required under the Administrative Procedure Act. Trying to coax state governments into mandating “nonbinding” federal policy does not inspire confidence that NHTSA is planning to play above board. Congress should take note.
II. Model State Policy
Avoiding a patchwork of state automated vehicle laws and regulations remains one of the most challenging elements of this space. CEI has long called for the development of reasonable state model legislation, although we have criticized misguided efforts such as those undertaken by the Uniform Law Commission. While NHTSA’s guidance document stops short of actually developing model legislation, it does contain a number of solid state policy recommendations.
NHTSA begins this section by highlighting its authority relative to the states: “DOT strongly encourages States to allow DOT alone to regulate the performance of [highly automated vehicle] technology and vehicles…. The Vehicle Safety Act expressly preempts States from issuing any standard that regulates performance if that standard is not identical to an existing FMVSS regulating that same aspect of performance” (pp. 37-38).
This is clearly aimed at states such as California, which has been plagued by regulatory delays thanks to incompetent (although well-meaning) legislative drafting and poor regulatory implementation. California’s legislators wrongly believed their Department of Motor Vehicles was capable of serving a NHTSA-lite role, something that has been proven false.
In addition, NHTSA recommends that states undertake a legislative and regulatory audit of their motor vehicle and related administrative codes to look for conflicts between existing laws and regulations and automated vehicle technology. NHTSA has already conducted a similar preliminary audit of its FMVSS, which was released in March 2016.
Unfortunately, one key element left out was one CEI raised in 2015 before the Uniform Law Commission. In our comments, we noted that the Commission’s draft model policy failed to include test driver license reciprocity. If states with metropolitan areas that span across multiple states fail to recognize out-of-state licenses, they are unreasonably restricting the test driver labor pool, which increases costs and access to qualified test drivers, perhaps perversely reducing highway safety.
The District of Columbia has been struggling for several years with such a provision, so this represents an unacceptable omission from NHTSA. NHTSA’s recommendation that “[t]he operators testing the vehicles must hold a valid State driver’s license” (p. 43) should be clarified to cover a test driver licensed in any state.
III. NHTSA’s Current Regulatory Tools
This section covers NHTSA’s existing regulatory authority in matters of interpretation, exemption, rulemaking, and enforcement. A brief discussion of letters of interpretation and NHTSA’s FMVSS exemption powers is warranted before moving to the final section in which NHTSA gives its new authority wish-list.
Interpretation letters request guidance on current laws and regulations from NHTSA’s Office of General Counsel. For instance, Google earlier this year received an interpretation from NHTSA determining that its artificial intelligence system qualifies as a “driver” under FMVSS. Interpretations can be a much quicker way to bring new technologies to market, as changing an FMVSS can take years whereas an interpretation can occur within mere months.
Exemptions are more powerful than interpretations, as manufacturers are able to escape FMVSS altogether, rather than merely seek a novel alternative method of compliance. However, these are greatly restricted and currently serve few practical benefits to post-testing deployment.
Exemptions are generally limited to a 2,500 unit manufacturing run per year for two years, with a two-year renewal opportunity at the end of the period. A manufacturer able to show “substantial economic hardship” can produce up to 10,000 vehicles per year for three years, with a three-year renewal opportunity at the end of the period. Those seeking an exemption solely for testing may wish to consider the new Sec. 24404 replica exemption enacted under the FAST Act of 2015, although NHTSA notes in its guidance at endnote 68 that most vehicle designs currently being pursued by developers would likely not qualify.
IV. Modern Regulatory Tools
This entire section deserves a great deal of scrutiny, as it discusses changes in policy NHTSA is considering pursuing. Some of these potential changes are substantial and require significant changes to the authority granted to NHTSA by Congress. While many within the agency believe that additional power is all they need, Congress should be wary of any efforts to turn 50 years of auto safety regulatory policy on its head.
Since the Vehicle Safety Act was enacted in 1966, NHTSA has required manufacturers to self-certify their vehicles as meeting FMVSS. Self-certification is the process that continues to this day, which is augmented by post-market NHTSA compliance testing. In contrast, the European Union operates under a type approval regulatory authority. Type approval is a form of pre-market approval whereby the regulator receives a vehicle from the manufacturer to test. It requires substantial additional government resources, and is significantly more intrusive and time consuming.
NHTSA’s guidance contemplates two forms of pre-market regulatory authority, one that would replace self-certification and another that would augment it. Neither of these suggested mechanisms should be authorized by Congress.
The case for pre-market approval made in the past by proponents is that the U.S. isolates itself from the export market by defying the “norm” of type approval. However, as the U.S. is the world’s largest motor vehicle market, this cuts both ways. Granting NHTSA pre-market approval authority will work against bringing U.S. developers’ automated vehicles to the U.S. market—even if it speeds approval of those U.S.-produced vehicles in other countries.
One sensible thing NHTSA does recommend in this section is expanding exemption authority (p. 76). This could be done by simple amendment by Congress, increasing the annual manufacturing limit (for example, from 2,500 to 5,000 vehicles) and extending the exemption period (for example, from two to five years). This could increase the volume of early-generation automated vehicles deployed to consumers without upsetting the apple cart. Of course, eventually NHTSA must undertake a wholesale reevaluation of its FMVSS regime, but this should be done in a thoughtful manner and it is premature to attempt such dramatic reform at this time.
CEI will expand on these and other elements of the guidance document in formal comments to the agency. The upshot is that many of these nonbinding recommendations are welcome, but NHTSA must work to limit its precautionary impulses, which have the potential to delay and increase the cost of automated vehicle deployment—meaning more preventable crashes, more injuries, and more deaths.