Last week in the city of Manchester, U.K., a forty foot sink hole appeared in one of its main roads. As sink holes go, it was big, but globally nowhere near record size. Fortunately there were no autonomous vehicles in the vicinity and all the 'normal' vehicles avoided falling into it.
Go forward a few years to a time when the fleet is largely autonomous and it could have been a major incident because it is difficult, if not impossible, to see how an autonomous vehicle could be made to detect the absence of road surface that was the sink hole in sufficient time to stop safely if sensible road speeds are to be achieved. Slowing them all down to walking pace is hardly the answer. In short, it could have resulted in significant loss of life, especially if it had been flooded, which is highly likely.
Perhaps it is due to my vehicle research background, but it would not be too much of an exaggeration to say that I have a list as long as my arm of reasons why I would never want to ride in one of these vehicles, or even be in the vicinity of one when it is on the move. I would feel a lot safer riding in a pilotless aircraft, though that isn't saying much.
The current enthusiasm for autonomous vehicles makes me wonder whether any of their proponents have even seen, let alone read, Nicholas Nassim Taleb's The Black Swan. If not, perhaps they should, it could save a lot of expense developing something that is, to me at least, glaringly unworkable.
Now, if we were discussing the application of the driverless technology to the provision of driver assistance, then we would be on to something, both possible and desirable with one pair of eyeballs mark 1 in the loop to act as a safety net.