May 13, 2019

Autonomy and the Law

Perrone Robotics recently hosted University of Virginia School of Law’s “Common Law” podcast team and our CLO Mike Raschid participated in the discussion that examined the current and future state of the law governing autonomous vehicles. As part of the discussion, the show’s hosts, Dean Risa Goluboff and Vice Dean Leslie Kendrick, visited our site and took a ride in one of our test vehicles.

The podcast noted that the emergence of autonomous vehicles will require complicated changes to our torts and products liability legal regimes and those changes will not be made overnight. For example, the podcast discussed the widely-held belief that as autonomous vehicles become more prevalent, the number of accidents will drop dramatically. But questions remain as to how liability will allocated when roads are shared by autonomous and non-autonomous vehicles, and also when there are significant numbers of autonomous vehicles on the roads. 

Professor Kenneth Abraham of UVA School of Law, a leading expert on the tort and insurance system, suggests that a “strict liability” system will save time and resources. In this system, a manufacturer is responsible for all accidents that “arise out of the operation of an autonomous vehicle”. Those involved in an accident will not go through the costly and complex process of showing fault and attributing such fault to a particular product defect or party that caused the accident. Rather, the liability system will allocate payments based on the data from the accident.

While the vast majority of auto accident cases are settled out of court, this new legal and insurance regime proposed by Professor Abraham would be a welcome change – and further reduce the likelihood and incidence of litigation arising from auto accidents.  What interests us at Perrone Robotics is how the availability of information and data regarding accidents will grow substantially when autonomous cars are introduced in large numbers on public roads. Currently, vehicles collect data such as vehicle speed, brake status, throttle position, ignition cycles, delta-V, and seat belt status. This is a big leap forward over previous – and still used – techniques such as skid mark analysis and fender crumple analysis. However, is still information about one car only.

In the coming years where autonomous vehicles deploy in larger numbers, there will be a wealth of data available for very clear analysis of what happened in any situation. That data will be useful for years to come in accident reconstruction while the new legal and insurance regimes emerge. Multiple sensors will collect internal and external environmental data, vehicle speed, position, other vehicles or obstacles nearby, and desired and actual trajectory to a very granular level. These vehicles will also communicate with the road and highway infrastructure to track changes over time. Thus it will be very clear what is happening inside and outside the vehicles involved in an accident.

Accidents will continue to happen.  While autonomous cars will be as safe as possible, the laws of physics still apply. If someone steps out in front of a moving vehicle it may not be able to stop or swerve in time (although it will do its best and that should minimize damage/injury).  But with the vast amounts of computing power, data-gathering and self-checking that are involved in the operation of an autonomous vehicle, we’ll all be able to more safely share the road and our environment with such vehicles.  Finally, should an accident happen even with the safety features of an autonomous vehicle, we will have a lot of data to determine how the accident occurred, and how it might be prevented in the future.