First DARPA Grand Challenge

Background

The DARPA Grand Challenge was a prize competition for driverless ground vehicles, run by DARPA, the research agency of the U.S. Department of Defense. The rule was blunt: the first team to pass the qualification tests and then complete the desert course in under ten hours would receive USD 1 million ^[1]. That single condition set the character of the event. A vehicle had to be fast, reliable and fully autonomous at the same time, because anything that stopped, stalled or needed a person to take over would run out of time.

What happened

The first running took place on 13 March 2004 in the desert outside Barstow, California ^[1][2]. Fifteen vehicles left a starting gate at dawn ^[1]. More than 100 teams registered in that first year ^[2], so the field that actually lined up at the gate was a small part of the groups that signed up.

The route ran across open desert toward the Nevada state line. DARPA's own retrospect describes a 142-mile course running across the desert to Primm, Nevada ^[1]. Wikipedia gives the length as 150 miles (240 km) ^[2]. The two accounts differ on the exact figure, but agree on the scale: hours of driving over rough, unmapped ground with nobody at the wheel.

Result

No vehicle finished the course. DARPA says the top-scoring vehicle travelled only 7.5 miles ^[1], while Wikipedia gives the best distance as 11.78 km (7.32 mi) ^[2]. Under either figure the leading machine stopped a long way short of the finish. No winner was declared and the USD 1 million prize was not awarded ^[1][2].

Why it mattered

The value of the event lay in what it made visible. Before the race, autonomous driving could be shown in short demonstrations on prepared ground; a fixed desert course with a clock running turned it into a measurable test that anyone could compare against. The outcome — a field of fifteen vehicles, none of them able to complete the route — set a clear baseline for how far the technology still had to go, and gave later attempts something concrete to beat.

The response was immediate. DARPA announced a second Grand Challenge one day after the first ended, for the fall of 2005 ^[1]. Rather than treating the result as a failure and dropping the programme, the agency scheduled another attempt, keeping the same basic format of a desert course, a time limit and a large cash prize. That decision turned a single disappointing race into the start of a series.