The path to hell is always paved with good intentions (The government is your friend.)

The comment by Karl Crary (scroll down to his comment on March 19, 2004) shows that irrespective to the many tests one can imagine, you are never in a condition to test all your variables, especially when it comes to an autonomous vehicle. In effect as Karl says " In the race, DARPA gave them a course not much wider than the vehicle itself, so the error of the GPS was a significant factor. (This was apparently in an attempt to be helpful: the course was narrow enough to miss all major obstacles.) This revealed a bug that had never come out in testing: they did not compensate properly for the error in GPS. As a result, Sandstorm found itself adhering tightly to a course a few meters off from where it was supposed to be. The vehicle correctly identified the obstacle that killed it, but thought that it was not allowed to steer around it." the precaution DARPA took killed the vehicle's ability to do real autonomous decision making. Even if Sandstorm used the services of a DGPS quality inducing a cm difference between its known position and the position given by the GPS, if there was no correction made on the offset between the true position and the one given by the GPS there was no way to eventually make it to the end of the race. When you look at the video you see there is no good reason as to why Sandstorm should have failed in this part of the race....