I agree; as long as they study it, eventually they'll come up with solutions. But I'm glad they're finally realistically painting the challenges and acknowledging that computers really aren't faster than the human brain in regards to calculations.
That said, I find it interesting that mapping is such a big challenge. I think they're spending way too much resources and computing power to have a super-accurate 3-D map. They'll find that the solution is what people do all the time - take 'short cuts'. During most of my driving I know the general route - what direction, what streets to take, etc. But I don't memorize every feature about every inch out there - I generalize, or 'short cut' it. Ex: I know the general wide of the roads, I know approximately where curves are, etc. Basically like a GPS map. That should be their starting point. Then program common traffic knowledge - what an intersection 'looks' like, what a traffic light looks like and where 'red' 'should' be (I assume there's state laws regarding this) - so the computer shouldn't have to recognize red, just where it's placement on a traffic light 'should' be.
I applause their determination, innovation, and drive (pun intended); but I think they should alter their goal to improve computer-aided driving first and through that process evolve it into more and more computer-led driving.
Oh, and back to the mapping issue - I'm betting that along the way they'll decide to let the individual vehicles do the scanning for them. I mean, they're the ones on the road. Attach additional sensors that would transmit that information back to Google to keep the maps up-to-date. Plus, they'd sell it as - 'hey! If an accident occurs on your route, cars would transmit that information to Google and Google could then send that information to vehicles in the vicinity so they could develop rerouting plans, etc. ala the cellphone strategy Batman employed in The Dark Knight.
Which would (or rather, should) raise all sorts of privacy issues, but as the articles shows, they're still a ways off.