What happened
Bounded success. The second event achieved the specified demonstration, with multiple finishers rather than a lone exceptional entrant. This is success at autonomous desert traversal; it is not evidence that the same systems were ready for unrestricted public-road service.
- No vehicle finished in 2004. Five finished in 2005, when Stanford won the $2 million award. Grand Challenge
- Registrations rose from 107 in 2004 to 195 in 2005; smaller numbers qualified for the races themselves. Stanley: The robot that won the DARPA Grand Challenge
- The technical paper attributes Stanley’s performance to an integrated development effort involving perception, planning, control, machine learning, testing, and several institutional partners. Stanley: The robot that won the DARPA Grand Challenge
What might explain it
Plausible
A repeatable public test may coordinate complementary engineering work around a shared definition of progress. Grand Challenge Stanley: The robot that won the DARPA Grand Challenge
Alternative explanation: Extra development time, accumulated research, and improved hardware could explain improvement even without a competition.
Plausible
Retaining the challenge after failure may allow learning that a one-shot award would lose. Stanley: The robot that won the DARPA Grand Challenge
Alternative explanation: The second event changed entrants, purse, and course; it is not a controlled comparison of persistence alone.
Questions to take into practice
- Before repeating a challenge, ask whether the first attempt revealed tractable bottlenecks or an unrealistic target.
- Preserve evaluation data and technical accounts so failure can inform the next round.
What remains uncertain
- The second round changed the purse, course and entrants while allowing more development time, so the improvement cannot be attributed to the larger reward alone. The costs and learning of unsuccessful entrants are also needed to assess the program’s wider value.