Do you know none of that has anything to do with the issue at hand (the OP lauding Google for bringing successful products to market, while they didn't even bring them to market yet?)
Although DARPA rules require that the team have an American representative, teams like Team Berlin [1] and CarOLO [2] were run by German universities. I certainly agree that universities produced a lot more entries than the vehicle industry did - but I don't know if that means Mercedes and Toyota had secrets they didn't want to give away in 2005, or if it means they didn't have anything worth showing off.
Of course, we'll know much more when companies actually start bringing these products to market, and we shouldn't count Google's chickens before they've hatched. But by my reckoning they got a pretty good head start - I certainly wouldn't bet against them.
[1] http://archive.darpa.mil/grandchallenge/Teams/Teamberlin.asp [2] http://archive.darpa.mil/grandchallenge/Teams/CarOLO.asp
And did you know the Grand Challenge winning cars made use of LIDAR and machine vision to identify road features and obstacles, high precision GPS, and inertial navigation? And that very similar sensors can be seen on Google Street View?
If you wanted a test data set, to refine your self-driving car by simulating driving every road in the developed world, you could scarcely do better than Google's data. This could be valuable in advancing from a car that works well on wide, well maintained freeway to a car that can work well around towns and cities.
[1] http://www.youtube.com/watch?v=uoiJeIb0wBA#t=21m50s [2] http://en.wikipedia.org/wiki/Sebastian_Thrun