NHTSA and Autonomous Vehicles (Part 3): Hearings and Strange Bedfellows


This is the final segment in a three-part series on NHTSA and autonomous vehicles.  The first two parts can be read here and here.

So what went down at NHTSA’s two public hearings?  I could not find video of the first hearing, which was held in Washington DC, and so I’ve relied on press reports of the goings-on at that initial hearing.  The full video of the second hearing, which was held in Silicon Valley, is available on YouTube.

Most of the speakers at these two hearings were representatives of tech and automotive industry companies, trade organizations, and disability advocacy groups who touted the promise and benefits that AV technologies will bring.  Already, vehicles with automated features have a level of situational awareness that even the most alert human driver could never hope to match.  Sensors and cameras can detect everything that is going on around the vehicle in every direction–and AI systems can ‘focus’ on all that information more-or-less simultaneously.  Human drivers, by contrast, have a limited field of vision and have trouble maintaining awareness of everything that is going on even in that narrow field.

AI drivers also won’t get drunk, get tired, or text while driving.  (Well, actually they could send texts while driving, but unlike with humans, doing so would not hinder their ability to safely operate a vehicle).  Their reaction time can make human drivers look like sloths.  Perhaps most significantly, they could give people with physical disabilities the ability to commute and travel without the need to rely on other people to drive them.  If you follow developments in the field, then all of that is old news–but that does not make it any less enticing.

But the hearings also had their share of expressions of concern.  Some speakers voiced outright opposition to the “fast-tracking” of regulations that would allow the widespread deployment of AVs.  Consumer watchdog groups, such as (wait for it) Consumer Watchdog, claimed that companies were pushing for regulatory approval of AVs before the technology was fully understood and ready for deployment.  These consumer groups may have found an unlikely ally in the form of the Association of Global Automakers, an industry group that represents five of the largest Asian automakers (Honda, Nissan, Toyota, Hyundai, and Kia).  An AGA representative warned NHTSA at the first hearing to “not bind itself to arbitrary, self-imposed deadlines” and “instead consider the development” of such regulations “incrementally.”  (Not coincidentally, most of the manufacturers represented by the AGA apparently have lagged behind both rival automakers and tech companies in developing AV technology.)

But perhaps the most commonly expressed concern during the NHTSA hearings–particularly the second hearing in Silicon Valley–came from a quite different direction. These individuals expressed concern  Specifically, many representatives warned that absent decisive federal regulatory action, the regulation of autonomous vehicles might devolve into a “patchwork” (a word that at least 4 speakers used at the Silicon Valley hearing) of different state laws.  If those state laws deviated in any significant ways, the resulting Patchwork would make it virtually impossible for manufacturers to mass-produce vehicles that could be sold throughout the country.

The Patchwork is a concern because of a quirk of American law.  By and large, the issue of what can be driven on a road is a matter of federal law (through NHTSA, FMCSA, and an alphabet soup of other Department of Transportation agencies), while the issue of who can drive on a road is a matter of state law (in the form of laws governing the licensing of drivers).  This never has a been a problem before because there is a clear distinction between the vehicle’s equipment and the vehicle’s operator in traditional, human-operated motor vehicles.  But as automation increases, the distinction between “operator” and “equipment” becomes blurry–and in case of a fully autonomous vehicle such as Google’s, it is almost meaningless.  This means that there is plenty of space for states to pass regulations governing what computers can and cannot do in terms of vehicle operation–thus giving rise to the possibility of an unmanageable Patchwork.

Fortunately, the US Constitution has this wonderful thing called the Supremacy Clause that makes state laws subordinate to federal law, including regulations passed by federal agencies such as NHTSA.  One corollary to the Supremacy Clause is that when a federal agency creates a comprehensive regulatory scheme–such as NHTSA’s Federal Motor Vehicle Safety Standards–that scheme preempts state laws that cover the same subject matter. NHTSA can take this one step further, because a federal statute specifically states that states can create motor vehicle safety standards only if they are identical to the federal standards on the same subject.  So if NHTSA decides to promulgate a comprehensive set of regulations stating what AVs and their AI ‘drivers’ can and cannot do, that would ensure uniformity for AVs under US law and prevent the Patchwork from becoming reality.

This has created a fascinating dynamic–usually industry groups oppose new federal regulations while consumer groups support them.  In the AV world, however, consumer groups are basically pushing for NHTSA to sit on its hands while automakers and tech companies are, for the most part, pushing NHTSA to crank out new regulations ASAP.  NHTSA has promised to issue initial guidelines by the end of the summer, but those guidelines won’t have the preemptive effect needed to stop the Patchwork unless they are accompanied by the laborious formal rulemaking process.  It will be interesting to see what happens over the next few months and years, particularly since the imminent election could lead to leadership changes at NHTSA and DOT that could affect both the timing and content of AV regulations.


Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.