NASA's AI drone take on human drone pilot in race

25 November 2017

Researchers at NASA's Jet Propulsion Lab (JPL) in Pasadena, California, revealed Tuesday the results of two years spent on the development of alogrithms for autonomous drones using technology also used for spacecraft navigation, funded by Google.

The race, held on 12 October, pitted NASA drone pilot Ken Loo against custom-built drones named after comic book characters Batman, Joker and Nightwing, that could reach speeds of up to 80 miles per hour. The robots held the advantage in the race, thanks to factors such as natural human aggression and fatigue.

"We pitted our algorithms against a human, who flies a lot more by feel," Newsweek reported Rob Reid, the project's task manager, as saying. "You can actually see that the AI flies the drone smoothly around the course, whereas human pilots tend to accelerate aggressively, so their path is jerkier."

''This is definitely the densest track I've ever flown,'' Loo said in a statement. ''One of my faults as a pilot is I get tired easily. When I get mentally fatigued, I start to get lost, even if I've flown the course 10 times.''

The drones weaved through the twists and turns of the warehouse obstacle course hitting speeds as high as 40 miles per hour.

Loo operated his drone from a set of controls to fly faster, taking risks with aerial stunts and aggressive movements. The AI designed by NASA's Jet Propulsion Laboratory and funded by Google, was not flashy, but it displayed smoother, consistent movement, always hitting the same racing line every lap.

The AI was more accurate and took fewer risks, but the more cautious flights cost the computer precious time and after dozens of laps, Loo emerged exhausted but victorious. Averaging 11.1 seconds per lap, he beat the autonomous drone, which averaged 13.9 seconds.

 search domain-b