Self-driving cars are legal in the United States. They've always been legal, because they've never been outlawed. This is America, after all, where if no one says you can't do it, you can totally do it.
But as automakers and others close in on making the technology commercially available---cars with at least some autonomous capabilities could be on the road within three to five years---such de facto legality is hardly reassuring. Eventually, state and federal policymakers will have to come up with rules. That's fine, because everyone in this game wants rules, so they know the guidelines they're playing under, from an early stage. If you develop a system using radar and cameras, then the feds later demand you include LIDaR, you're out of luck.
The problem is, policymakers are poorly equipped to decide just how to regulate robo-cars. The tech is new, complicated, and being developed in different ways by different companies, so just understanding how it works is difficult, let alone knowing how to make it work safely.
But this technology is charging ahead, despite the lack of a coherent national policy, and a patchwork of rules from the few states that have legislation on the books. Some companies will opt to ask forgiveness later rather than permission now. They're simply moving forward. "That may force very difficult questions about what the legal regulation of the driver should be," says Bryant Walker Smith, an assistant professor at the University of South Carolina School of Law and affiliate scholar at the Center for Internet and Society, who studies self-driving vehicles. "It will be driven by what actually gets to market."
In other words, it may be the Googles and Delphis and Audis of the world who play the deciding role in how this technology rolls out.
The National Highway Traffic Safety Administration (NHTSA) has addressed the question of automated vehicles, largely by announcing, almost two years ago, that it is studying the issue and will decide "whether it should encourage and/or require application of the most promising crash avoidance technologies through regulation." It hasn't said when that decision will come.
"Washington should not be an impediment to technology improving our overall system," Secretary of Transportation Anthony Foxx said this week. "We want to encourage the work that's going on here, we want to assist in the innovation that is underway."
In the meantime, Nevada, California, Michigan, Florida, and Washington, D.C. have adopted laws regulating how the technology is tested and sold. The problem is these rules are inconsistent. Nevada, for example, requires a special license and registration, but that only applies to vehicles sold in the state. Florida's law is basically pointless, saying the state “does not prohibit or specifically regulate the testing or operation of autonomous technology.”
In California, the technology can only be tested and is not approved for consumer use. According to Stanford Law School’s Center for Internet and Society, 14 more states are working on regulations, and another dozen have voted them down. The states without rules could crack down on a self-driving car if they wanted to, using laws probating reckless driving, for example, to revoke registrations or refuse to register cars going forward. (New York has a law requiring a driver to keep one hand on the wheel at all times, but that doesn't mean the car can't drive itself.)
Automakers loath this patchwork approach and would much rather see uniform standards. Such standards, Walker Smith says, would provide a definition of autonomous vehicle safety, and a way to measure it. "It’s good to have clear regulations, on what to do to test those systems safely," says Audi engineer Daniel Lipinski, who helped develop Audi's autonomous car. Sean Walters, director of compliance and regulatory affairs at Daimler, which just introduced an autonomous 18-wheeler, agrees: "National standards are critical to the trucking industry, especially with respect to new and innovative technologies."
That said, there is some concern that government regulators who don't fully understand the technology---let alone the vision to see where it might lead---will muck it up. "We see a danger of actions taken too early to govern piloted driving 10 to 20 years into the future," says Audi spokesman Brad Stertz. "Such laws would have little application to initial levels of piloted driving that will emerge over the next few years. They could also jeopardize robust research needs." Google's on the same page. "It's really hard to try and anticipate how the technology might be used in the future and write laws for every eventuality. We think policymakers should learn about the technology and see how people want to use it first before putting a ceiling on innovation," a spokesperson says.
That's because---and this is the crux of the problem---the government is poorly poised to regulate this technology. US Secretary of Transportation Anthony Foxx admits it's hard "for a regulatory agency to keep pace with all the changes that are happening so rapidly ... We've just got to continue working to keep up with it." Compounding this problem is a catch-22: We can't gauge the safety of these cars without testing them on the road, and we don't want them on the road until we know they're safe. Foxx compares this problem with regulations governing UAVs, saying states should keep "opening the nets" a bit to allow testing on public roads.
Beyond keeping pace, there's the fact that autonomous technology muddles the classic jurisdictions of state and federal regulation. As it works now, feds control the technology going into cars---like airbags and seat belts---while the states dictate the operation of those vehicles by establishing traffic regulations. A self-driving car obliterates those distinctions with hardware and software that controls how the car drives. So it's not even clear who should be writing the rules, let alone what they should be.
A third problem: NHTSA is a reactive, not proactive, regulator. "Historically, they don't get out in front of new technologies," says Steve Shladover, a University of California, Berkeley transportation expert helping the state DMV work with the technology. "They respond to things that come into the vehicle market," making sure they're safe, and often eventually requiring them.
In the past, the agency's tried to get ahead of technology and been slapped down. In the early 1970s, it wanted to require ignition interlocks to prevent cars from starting if those inside weren't buckled up. People hated the idea, and Congress made such a mandate illegal. But it hasn't totally given up on being proactive: It's working on getting ahead on vehicle-to-vehicle communication, which uses a dedicated short range radio network to let cars communicate things like speed and position to each other, to improve safety. Foxx says NHTSA should have a proposed rule requiring the technology by the end of this year, and a final rule in place by the end of 2016. It may want to get ahead on self-driving cars, too.
Whatever the feds and states decide to do, self-driving cars are coming, and soon. Google, Audi, Mercedes-Benz, and Nissan promise to have cars with some level of autonomous within a few years. Tesla plans to offer a more limited version of autonomy next month via an over-the-air software update. (A company rep says "we continue to engage with regulators throughout our process of development.")
"The real change will be driven by specific companies with specific plans," says Walker Smith. As these technologies hit the market, regulators will be able to assess them, determine if they're safe, and make rules accordingly.
Just like it's always been done.
