I think once they get to 'as good as people' we should let them on the road, because they are 'as good as people' but never get tired, angry, and/or drunk :)
As good as humans implies that over one million people would be killed by robots every year. Some of the families, loved ones and attorneys of these people would be able to make very convincing cases that they would not have died if a human were driving. This is a hard problem to overcome.
However, that's horrific and implies that the engineers are using something of an evolutionary algorithm. Obviously, this does not cut the cheese.
IF the sensors would be standard maybe we could create fake inputs and test different situation in a simulation, have some basic unit tests, now I am afraid that this companies are just tweaking things, something that today works tomorrow after an update may not work.
Even if the NN evolve if it recognizes a bike in image X it should still recognize it after it learns more or is updated with new hardware that supports more neurons.
We need a better measurement that would not involve killing people, at least as a lower bar before accepting this cars for testing.