October 5, 2018

Three Cheers for Autonomous Vehicle Policy Guidance v3.0



Yesterday, the Department of Transportation (DOT) released the third version of its guidance for the autonomous vehicle industry. In its write-up of the release, The Verge said the plan is “intended to open the floodgates for fully driverless cars.” Autonomous vehicles will be a boon to safety and efficiency in our largest transportation sector. We welcome the deluge.

The changes in guidance likely have the most significant implications for the trucking industry, as noted by Bloomberg:

The U.S. Transportation Department has given a boost to companies working on automated long-haul trucks, saying an artificial intelligence system could constitute a “driver” under federal trucking rules in a bid to ease barriers to the technology.

Long-haul trucking represents some of the lowest-hanging fruit for automation because highway driving is relatively easy for artificial intelligence, boring for humans, and costly for trucking companies (given the labor costs).

In its guidance, the DOT also signaled that it is willing to preempt state and local laws to create a consistent regulatory environment for autonomous vehicles. Given that automotive transportation is almost inherently interjurisdictional, it is critical for the federal government to prevent a patchwork of regulations from stifling innovation and slowing the development of this breakthrough technology.

Among the most promising inclusions in the guidance, the DOT explicitly notes that it will no longer assume that the driver of a vehicle is a human (or that a human is even present onboard). This is a welcome development, and one that the Niskanen Center emphasized in recent comments to the National Highway Traffic Safety Administration (NHTSA). In particular, we argued that while “the definition of a vehicle ‘operator’ or ‘driver’ needs to be updated to reflect non-human operations in an autonomous vehicle,” NHTSA will also need to consider how it intends to define the scope of what constitutes a non-human operator.

Settling on a clear technical definition of, for example, an artificial intelligence, neural network, or machine-learning algorithm capable of safely and effectively operating a motor vehicle is no small task. Nonetheless, this definition will undoubtedly be a cornerstone of any future regulatory regime governing this technology’s deployment, and will likely be the deciding factor in whether and how autonomous vehicles can be successfully integrated on American roadways.

Luckily, this new guidance also lays a strategic roadmap for how the DOT plans on addressing some of these questions, beginning with the clear need to update legacy regulations that pose a barrier to the research, testing, and ultimate deployment of these systems. For example, the new guidance announced a number of forthcoming rulemakings, including one that would consider how to exempt autonomous vehicles from certain standards that only apply to human-operated vehicles.

The guidance also drops the 10 self-drive car sites picked by the Obama administration and expands the pilot program more broadly (mimicking the Trump administration’s drone pilot program). Given the potential of this technology to prevent tens of thousands of deaths each year, we applaud the administration’s push to accelerate testing nationwide.

This clear and non-binding guidance from DOT is a paradigmatic example of the “soft law” governance approach we have championed for all emerging technologies. As we concluded in our regulatory comments submitted to the Department of Transportation earlier this year:

Autonomous vehicles are no longer the purview of science fiction; their advent is rapidly approaching. The sooner we can begin clearing the roadblocks to their deployment, the sooner all Americans can benefit from safer streets.

With the release of updated guidance on how the DOT intends to treat autonomous vehicles, we are one step closer to that safer, more autonomous future.