From a technology perspective, I think there are three parts to what’s happening here:
A single autonomous car driving down the street and not hitting anything – this is what most of the attention is on now.
Optimizing traffic flows when some or all cars on any given road are autonomous (two very different problems) – this is what we really see in the videos above.
Optimizing a fleet of autonomous on-demand cars in a city on a real-time basis. Where are the cars now and where do you want them to be in 17 and 34 minutes to minimize traffic, trip times and wait times?
Of these, the second and third feel much more like Google or Uber challenges than Apple or (perhaps) Tesla challenges. These are questions of algorithms and large scale computing systems, not design, experience or ease of use. They are also why maps have become so important – maps are PageRank for the real world. For car companies maps used to be an accessory no different in strategic terms to the CD player, but now they’re a site of existential worry.