I don't have any credentials to speak on the subject, but it seems that robotic cars is an area where traditional startup advices along the lines of "move fast" has to be taken very cautiously. I also believe that Google being paranoid about first robotic car killing someone is very much justified. Rightly so.
In any case, all of this will happen eventually. We will have robot cars on the road eventually, and eventually one of them will get into some sort of accident with big publicity potential. Any company involved in the industry probably ought to be managing what happens to them when that happens.
Especially if, Ford Pinto style, they've publicly stated that [for financial/reputational reasons] they shouldn't be as concerned about safety as another market participant, or that they should should aim to be "only just" better than fallible human drivers who frequently are held accountable for their fallibility.
I wouldn't want to be on the design team of a robotic car the first time it hits a child, even if the child acted in a reckless and unpredictable way that even a human mind would have been unlikely to anticipate and the software had demonstrated a far superior safety record to human drivers overall.
The model case is probably the ongoing Toyota "unintended acceleration" case, in which it appears that poor software engineering really did contribute to fatal accidents. It's likely to cost Toyota a lot but not reach through to the design team.