Autonomous vehicle crashed….by human
Testing the software of Cruise Automation, a startup developing an autopilot system, a self-driving car crashed. However it did not crash while driving by itself, rather it crashed just after the handover from the automatic system to the human driver.
This points out one of the main problems with semi-autonomous systems, which is bringing a driver back into control after the computer has been driving. The accident happened while doing tests in the city of San Francisco, by far more challenging than most of the highway tests being conducted by most companies.
Law professor Bryant Walker Smith (University of South Carolina ) called this a “mushy middle’ of autonomy. Prof. Smith wrote a book-length paper on the legality of autonomous vehicles. Engineers at Ford Motor have been skeptical of systems that can switch back and forth between humans and computers, rather they prefer systems which take full control like Google’s efforts to the system of Cruise Automation that was installed on an electric Nissan Leaf.
Cruise Automation claim to fame is a retrofit system for late-model Audi A4 and S4 sedans called RP-1 that could be purchased for $10,000 and was supposed to be available in 2015 in a limited quantity. That system was meant for highways, not city environments, and different from the one the Nissan Leaf was equipped with.