The Road to Automated Vehicles Could Be Bumpy, Slow

January 22, 2018 by

Driving automation promises to greatly improve the safety of human mobility not only for occupants of motor vehicles but for pedestrians, cyclists and other road users. Current headlines, however, likely overestimate the speed with which this transformation will occur and underplay the challenges that will arise when the earliest automated vehicles share the road with humans. The earliest implementations of automated driving likely will be limited, curtailing the expected safety improvements. Fortunately, a more deliberate evolution of the technology than the sudden transformation heralded in headlines can better ensure it achieves its safety potential.

There is little doubt that automating vehicle control function creates safer vehicles. Research from the Insurance Institute for Highway Safety-Highway has shown that temporarily automating vehicle control in emergencies reduces crashes and injuries as illustrated by electronic stability control (ESC) or automatic emergency braking (AEB). ESC halves fatal single-vehicle crash risk and reduces fatal multiple-vehicle crash risk by 20 percent. AEB combined with forward collision warning cuts rear-end crashes by half. Even so, there are reasons to remain skeptical that further automating driving is certain to eliminate the nine-in-10 crashes due to human error.

As anybody who has ever worked with a computer knows, computers frequently fail to do the things they’re supposed to do, even when those tasks are far simpler than driving. Moreover, we don’t know how humans will use driving automation available to them.

Making matters more unpredictable, developers of higher levels of driving automation are now contemplating letting the technology bend the rules of safer driving to more smoothly integrate with human drivers. Allowing computers to make the same errors as humans is unlikely to improve safety. Recent announcements about General Motors’ Super Cruise and Audi’s Traffic Jam Pilot indicate that the inexorable march toward hands-free driving has already begun.

However, even if these innovations were on every vehicle, their impact on safety would be limited. An IIHS analysis of 2014 crash data indicated that automating interstate driving a la Super Cruise at most would eliminate 17 percent of crash injuries and 9 percent of crash deaths. Automating driving in traffic jams might eliminate 15 percent of crash injuries and 2 percent of traffic fatalities.

These modest reductions could be fully realized only if the technology is 100 percent effective and if human drivers use the features, and there’s ample evidence that neither is true so the actual benefit will be less. Moreover, Super Cruise and Traffic Jam Pilot are currently only available as options on some of the most expensive vehicles.

The realities of the automobile marketplace may frustrate disruptive innovation proponents, but the inertia inherent in our nation’s complex system of roads and motor vehicles also provides an opportunity to monitor change and guide its development. State policies that companies testing automated driving systems on public roads report crash and mileage data to state authorities are prudent measures. Similar consideration should be given at the federal level.

It will take time to realize a safer and automated transportation system, so proven measures to improve safety cannot be ignored. Lower speed limits, enforcing safety belt use laws and laws against impaired driving, safer road designs could be used to reduce the crash toll while we await the benefits of driving automation.

Zuby is executive vice president and chief research officer for the Insurance Institute for Highway Safety.