After deep-pockets law suits nearly killed US general aviation manufacturing some sensible limited-liability laws revived it. So, IMHO this troublesome aspect of autonomous driving will get sorted quickly. Insurers will love insuring people when the accident rate goes down as a result of automation. (Believe me. Their premiums will not drop proportionately.) Autonomous driving has to be very appealing to anyone who cares about human life. And certainly to anyone who trades on its fragility. Thirty thousand people die a year on our US roads . Robots are already acquitting themselves very well on everyday roads. Even today's robots would do much better than people under highway driving conditions and in light traffic. And I do not think it long before all but the most challenging driving tasks will be managed routinely. Frankly, the 'bots could hardly do worse than people. It is a bloodbath out there.
This pop sci article has what appears to be a very factual update on the Google autonomous car project. (500,000 miles without a crash.) But most closed beta test drivers still take over during tricky maneuvers and on side streets. I answered the Slashdot poll and thought about twenty years before we were all automated (I should live so long). But I expect to see a lot of automated cars in ten years. Maybe even Google's automated taxi service in some locales... Maybe.