Researchers have been improving the ability of drones to fly autonomously for several years now, but these self-navigating drones had not mastered the skills to navigate the tight turns and complicated obstacles of the racing courses on which human drone-flyers compete. Until now.
Scientists at NASA’s Jet Propulsion Laboratory showed off a drone that can swoop and accelerate entirely with a camera to gather in environmental data, and artificial intelligence to parse it rapidly. They pitted the autonomous drone against a professional drone racer, Ken Loo, a Google engineer and Drone Racing League pilot.
After dozens of laps around the course, Loo proved the better pilot than the AI system, with an average lap time of 11.1 seconds, compared to 13.9 seconds for the computer-driven drone. But after hours of racing, Loo confessed he was fatigued, while the drone uttered no such complaint.
“This is definitely the densest track I’ve ever flown,” Loo said. “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 researchers plan to improve the drone’s navigation algorithm, and you can bet its performance will improve. They say the AI skills they’re designing to help autonomous drones navigate difficult spaces are not limited to racing. They could, for example, be applied to move goods around warehouses or even on the International Space Station.