Despite the seemingly unending wave of autonomous vehicle testing flooding the news, a lack of regulation and some conflicts in the current regulatory standards for vehicles could be holding the entire industry back. At very least, that is the viewpoint of several automakers, their legal representatives, and some individuals with historical ties to the regulatory branches involved, primarily the National Highway Traffic Safety Administration (NHTSA), in the setting of U.S. travel and vehicle standards. The problem, according to some, is that the need for new and rewritten regulation has not been taken seriously and that the process for changing the rules is both complex and expensive.
It is important to note that this particular problem is not unique to the advent of self-driving cars. Instead, according to Elliot Katz, a partner at McGuireWoods LLP and head of that firm's automated vehicle division, it's a problem of technology moving faster than regulation and has been "seen in a lot of industries." McGuire continues to say that the issue stems from the fact that "the rulemaking process is not a short one" and that it "is nothing short of labor intensive." Both time and effort also attribute to the cost of setting new rules or rewriting old ones. Adding to that problem, according to one former NHTSA Administrator, Joan Claybrook, is that the NHTSA simply doesn't have the appropriate budget or the appropriate staff to formulate new rules. Claybrook believes that the rules themselves will take at least five years to get in place. On the other hand, another former representative of the NHTSA, attorney Tim Goodman, isn't quite so optimistic and believes the process will take between seven and ten years.
The discrepancies in time frames come mostly from how complicated rewriting the rules is going to be. A report released by an extension of the of the U.S. Transportation Department in 2016, for example, pointed out 73 current standards that will make creating autonomy in vehicles difficult simply because they reference human drivers or specific human-vehicle interactions. Those are effectively forcing industry participants to focus on creating self-driving cars that still have a human driver behind the wheel. At least 30 of those standards will almost certainly cause compliance problems for automakers trying to reach Level 4 or Level 5 automation – which are used to effectively classify vehicles that won't need any human oversight at all to operate. Looking past the current regulatory standards, new rules need to be written that cover a huge range of categories. At the time of this writing, there are currently more than 900 pages of rules for human-operated vehicles on the books since the rules began to be written in 1967.
While some of those rules could be adjusted easily enough to suit both types of vehicles, new ones will also need to be devised. That includes rules for how testing should be performed, for example, and that alone presents a huge obstacle. A lot of that can be attributed to the difference between human intelligence and artificial intelligence. Humans tend to be extremely adaptable and although computer systems can learn to be more adaptable, developing ways to test for every possible scenario or eventuality an A.I. driven vehicle will encounter is a statistical impossibility. Any new regulation regarding the testing of autonomy in methods of travel will, as a result, require a vast amount of technical knowledge about how the systems work and what tests can be performed that will ensure systems are both generally safe and adaptable to untested situations. At this point automakers and self-driving tech firms are essentially on their own and testing in the dark while trying to apply the same or similar standards to what is already in place for human driven vehicles despite the differences in how the two systems "think" and react. Beyond that, humans present another unique problem because there are so many different voices involved across the various industries that will be affected by the new regulations and technology. By way of example, a piece of regulation already on the books pertaining to automatic braking system still has no specific standards about how the system's performance should even be evaluated. That is a direct result of disagreements about the methods they should be evaluated with and the standards to which the systems should be held.
That doesn't mean there have been no efforts made to efficiently and effectively begin to regulate self-driving autos, or at very least to get the process started. Back in July, for example, the Coalition for Future Mobility – which is comprised of, among other groups, automakers themselves – released a public statement in an attempt to pressure legislators to get the NHTSA to start working on the problems. The House committee later directly the agency to begin updating auto safety standards, allowing more self-driving vehicles under an exemption from some regulations while new ones are written. The NHTSA also has plans to release nonbinding guidance regarding safety in September, following an initial push from the previous administration during the prior year. Countries outside of the U.S., in the meantime, have already begun taking action, but some don't believe the current rate of change in the U.S. is enough. Claybrook, for one, does not think the matter has been "thought through" and the result if that doesn't change, is likely to be many more years of waiting for appropriate rules to be written before self-driving automobiles in the country can really hit their stride.