NHTSA and Autonomous Vehicles (Part 2): Will Regulations (Or Lack Thereof) Keep Automated Vehicle Development Stuck in Neutral?
This is part 2 of a series on NHTSA and Autonomous Vehicles. Part 1, published May 8, discussed the 5 levels of automation that NHTSA established, with Level 0 being a completely human controlled car and Level 4 being a vehicle that is capable of completely autonomous operation on the roads. Part 3 discusses NHTSA’s April 2016 public hearings on the subject.
I must confess that I am very much an optimist about the promise of Level 4 vehicles–and not just because I really, really love the idea of having the ability to do stuff on my commute to work without having to scramble for one of the 2 good seats on a Portland bus (yes, there are always only 2). The potential benefits that autonomous vehicles could bring are already well-publicized, so I won’t spend much time rehashing them here. Suffice it to say, in addition to the added convenience of AVs, such vehicles should prove to be far safer than vehicles controlled by human drivers and would provide persons with physical disabilities with a much greater ability to get around without having to rely on other people.
But while I am optimistic about the benefits of Level 4 vehicles, I am not optimistic that NHTSA–and NHTSA’s counterparts in other countries–will act quickly enough to ensure that Level 4 vehicles will be able to hit the road as soon as they could and should. As prior posts have noted, there are few federal regulations (i.e., rules that appear in the Federal Motor Vehicle Safety Standards) that would present a significant obstacle to vehicles with up to Level 3 automation. But going from Level 3 to Level 4 may present difficulties–especially if, as in the case of Google’s self-driving car, the vehicle is designed in a manner (e.g., without a steering wheel, foot brakes, or transmission stick) that makes it impossible for a human driver to take control of the vehicle.
The difficulty of changing regulations to allow Level 4 vehicles creates a risk that automated vehicle technology will be stuck at Level 2 and Level 3 for a long time–and that might be worse than the current mix of Level 0, Level 1, and ‘weak’ Level 2 vehicles that fill most of the developed world’s roads.
When a human is in essentially full control of the vehicle, as in the case of levels 0 and 1, the human drivers know that it is up to them to stay alert on the road. They might not always be alert, but at least they know they have no one to count on but themselves to ensure their safety.
But Level 2 technology can lull human drivers into a false sense of security–a vehicle with a combination of adaptive cruise control and lane keep assist will probably seem to most people to be “fully autonomous,” even though there are hazards that the vehicle is not designed to avoid. Even with Level 2 vehicles that have a full suite of automated highway driving features, human drivers might try to activate and rely on those features even in environments–such as along country roads–where the technology was not designed to perform. Manufacturers constantly remind drivers that they need to remain alert when Level 2 (and Level 1) automated features are active, but anecdotal evidence and common sense both suggest that humans generally discount such warnings. It seems like a safe bet that human drivers will try to let the automated features of Level 2 vehicles handle everything even when they would be better off taking control themselves (or at least acting as a second pair of eyes).
On the flip side of the coin, human drivers will likely try to take full control of Level 2 and Level 3 vehicles when it would be better if they let the automated system handle things. Level 2 and 3 vehicles will, by definition, have the capability of letting human drivers take control of the vehicle at a few seconds’ (or less) notice. The problem is that humans overestimate their competence at most tasks, and thus might seek to take control of the vehicle when, from a safety perspective, the automated systems would perform better.
For these reasons, it would be far better if the progress from Level 2 to Level 4 proceeded rapidly. But if NHTSA–and, as I understand it, most similar agencies in other countries–leaves its regulations as-is, fully autonomous vehicles could not be produced and sold in the US, thus leaving autonomous vehicle technology at a lower level of development. That has created a peculiar regulatory dynamic. Typically, companies worry about the corrosive effects of regulation on innovation and are skeptical of federal regulatory intervention. But in the AV sphere, most companies actually seem more worried about federal regulatory inaction and are pushing NHTSA to intervene and pass new regulations. The reasons why will be the subject of the next and final post in this series.