Stanley wins the DARPA Grand Challenge

In March 2004 DARPA offered a million dollars to any vehicle that could drive itself across 240 kilometres of Mojave desert. The best entry went twelve kilometres and caught fire. On 8 October 2005 five vehicles finished the second course, and the winner, Stanford's Stanley, a Volkswagen Touareg with five laser rangefinders and a camera, did it in six hours and fifty-three minutes.

Sebastian Thrun's team won by treating driving as a learning problem. The lasers could see only thirty metres ahead, too little for desert speeds; the camera could see further but could not judge the ground. Stanley learned, as it drove, which colours and textures in the camera image corresponded to the road its lasers had already confirmed, and used that to look ahead. Its speed was set by a model trained on how a human drove the same terrain. The probabilistic methods Thrun had developed for robot mapping ran throughout.

Thrun went to Google in 2007 and founded its self-driving car project, which became Waymo, and the Grand Challenge's participants populated the autonomous-vehicle industry. The competition is where the twenty-year project of the Stanford Cart and ALVINN turned into a race with money in it.