The world's most advanced robotic cars have come a long way in the last few years. Now they can imitate what a human does behind the wheel as they squeeze into parking spaces, display the flair of a taxi driver merging into traffic, and turn on their indicators before making turns. This improvement in 'autonomous vehicle technology' is partially the result of America's defense department insisting that improvements be made to meet its hope of one-third of U.S. ground vehicles being robotic by 2015. To meet that goal, the Defense Advanced Research Projects Agency (DARPA) has cut back on the old process of giving out large research grants and receiving nothing beneficial in return. Rather, it has started running a series of grand prix races for the robotic cars. The first of DARPA's Grand Challenges was unsuccessful in 2004, but the robot cars improved in 2005. The teams were given an extra year to prepare their vehicles because no Grand Challenge was held in 2006, so DARPA has made its Challenge more difficult. Entrants must keep to the tarmac and obey the road rules, and they also must avoid colliding with various other cars being navigated around the base by stunt drivers. Thirty-five teams are competing for only 20 spots in the race. Sponsorship has followed success for the teams that have done well in the race, such as 2005's first place finisher, Stanford University, and the runner-up, Carnegie Mellon University. Other academic entrants, as well as some ad hoc groups and corporate entries, also compete. The current mix of competitiveness and fair play will surely remain, and that seems to create for DARPA what the standard research projects have failed to generate.