If and when robot cars are widely available, I suspect that road accidents will be reduced only if human driving is banned. Else, your robot car cannot prevent some guys in a Hummer from plowing into you. Just design cars with no steering wheel.
So, what if not everyone can afford these fancy-shmancy cars? You want to prevent poor people from being able to drive to work, to school? Ah, we can subsidize them, just like healthcare right now. Maybe it's cheaper to force people to take public transit.
Don't get me wrong. I am getting old enough that driving is becoming a chore, and I stopped looking at cars with excitement more than 2 decades ago. Right now, I would not recognize a Tesla if it drives by me. OK, I may notice the quietness. But back to robot cars, I am afraid it may take a bit longer than people anticipate.
How long ago was it that Toyota throttle "unintended acceleration" was blamed on the poor drivers? We talked about that a lot on this forum. Then, finally when the controller software was determined to be in error and Toyota was fined $1.2B, there was not much public coverage. I started a thread here (
http://www.early-retirement.org/forums/f27/toyota-fined-1-2b-over-faulty-pedals-71162.html), but not too many people cared anymore. And it was just the throttle, for crying out loud.
It costs a lot of money to design and build autopilots for airplanes (avionics is a big part of the cost of airplanes), to test them and make them safe. The problem is more than having the software do the right thing in normal conditions. A fault-tolerant software must also recognize when one of its associated hardware or sensors has a failure or becomes flaky, and take proper corrective action. A robot car has to do a lot more, and I mean a lot, than the autoland autopilot I worked on 30 years ago.