In 2016, more than 37K people died in car accidents in the U.S. That same year, one person died in Florida when his Tesla’s Autopilot feature failed to sense a truck cutting across the road, though the death was ruled an accident since there was no system defect detected.
Therein lie the problems illuminated by the news that a woman was killed crossing the street when a driverless Uber car failed to detect the pedestrian. Once again, the question of who is at fault and could fatalities like these prove that autonomous vehicles won’t work in a world where humans will do unpredictable things have come to the surface.
What Can We Learn from the Uber Crash?
Driverless cars have been using public roads for almost a decade now. Between Waymo (testing driverless cars since 2009) and Google, driverless cars have already logged 8 million miles of on-the-road driving time and there are only 2 fatalities to show for it.
No fault has been assigned yet, but experts say that the technology – emergency stop – should be able to prevent fatalities like the one in Arizona this past March. Still, autonomous cars are believed to be safer than cars driven by humans.
Despite that belief, in light of the accident, Uber suspended its driverless tests in Tempe, Pittsburgh, San Francisco, and Toronto. Who is responsible and what should happen when accidents like these inevitably happen? There are 3 things that we can learn from this accident that could help guide the debate going forward.
#1: Lawmakers Have a Dilemma
Back in 2015, lawmakers in Arizona opened the doors for driverless experiments in the state. In a bid to attract investor dollars away from Silicon Valley they passed legislation that lifted regulations hindering companies like Uber and Waymo. In response to a non-fatal crash involving a driverless vehicle earlier in March 2017, Arizona’s governor balked at the idea of more regulation considering that the other driver, not the driverless vehicle, was at fault.
After watching the dashcam video of the fatality, he called for the suspension of the tests in Arizona and called on tech companies to make the technology safer. More than any other state in the country, Arizona has seen a record number of pedestrian fatalities since 2017. Lawmakers now have to decide whether or not to prioritize a business-friendly environment where these technologies can be tested and improved or to listen to the outcry from the public for greater regulation.
#2: Driverless Cars are Still A Ways Off
Driverless cars work using cameras and sensor technology. Lidar which is the technology Uber vehicles use, goes a step further and uses laser technology to get a better sense of what is around the vehicle in motion. Both fatalities due to driverless cars were found to have no defect and yet did not deploy the most basic safety feature required for autonomous vehicles – auto stop.
#3: Disengagement Numbers are Crucial to Improving Safety
Another point that lawmakers and investigators will weigh is the fact that there was a human passenger in the car capable of overriding the vehicle and stopping it to avoid the accident. That passenger appears to have been distracted. According to numbers reported out of California, of the 350K trips using autonomous vehicles with a human passenger, 63 times the human had to intervene (disengage).
Clearly until there is no need for human intervention, driverless vehicles will not be ready for the general public. However, figuring out the causes of disengagements and requiring those numbers, like California does, could go a long way to working out the problems with the technology.