So I went to see the DARPA Grand Challenge. I was able to talk with 6 or 7 of the teams and see their robots. One of the amazing things that I noticed was the simplicity of the robots. Most of the robots were modified off-road vehicles that had mechanical controls of the brake, gas and steering wheel. Then they had a couple of sensors in front, and a couple of really fast computers in the rear seat. Another point that stood out was the lack of stereo vision. I had thought that every team would have some kind of array of cameras set up so that they could do image processing and get range data. I only saw 2 teams with stereo cameras set up and one of those was a commercially availiable solution. The other teams I talked with said that they had tried it and were unable to get good results. Every team I saw was using 2D SICK Lidar sensors. I was kind of hoping that more teams would be using stereo vision, and would open up their source code. It is relatively inexpensive to get 2 webcams to do stereo vision, but lidar sensors cost ~4 grand each. While the completion of this race showed great advances in autonomous vehicles, there is still lots of work to be done.