Racing has always been a highly exciting sport for humans that demands instinctive reflexes. Lately, humans have also started racing the drones and now it seems that they won't be the only competitors in this race for a long time.
Researchers at NASA's Jet Propulsion Laboratory in Pasadena, California, recently executed this very thought. They raced drones controlled by a professional human pilot against an artificial intelligence (AI), timing laps through a slanting obstacle track. The race, which was conducted by the researchers on October 12, marked the ending of a two-year-long research about autonomous drones, which was sponsored by Google. As reported by NASA, the tech behemoth was interested in JPL's work with vision-based navigation for spacecraft, which can also be applied to drones. To demonstrate the progress of the team, JPL prepared a timed trial between their artificial intelligence and a top-notch drone pilot Ken Loo.
The team customised three drones, namely Batman, Joker and Nightwing, and also developed the complex algorithms that they required while flying at high speeds, avoiding obstacles. Google's Tango technology was also incorporated into these algorithms.
Researchers built the drones according to certain specifications and they were capable of easily going as fast as 80 mph in a straight line. However, on the obstacle racing track, arranged at JPL, they could manage only 30 to 40 mph speed before having to apply the brakes.
"We pitted our algorithms against a human, who flies a lot more by feel. 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," said Rob Reid of JPL, the project's task manager.
As a result, the AI drones flew more cautiously than Loo's ones, but consistently. While Loo's times differed more, the AI flew the same racing line in every lap. However, the algorithms of the drones are still undergoing tests, as they faced certain hiccups during the race. For example, sometimes the drones flew so fast that they lost their trajectories due to motion blur.
Loo, on the other hand, acquired higher speeds and was also able to perform impressive aerial corkscrews. However, he faced some limitations, such as exhaustion, which was never a problem for the AI pilots.
"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."
Following dozens of laps, Loo became more accustomed to the course and, as a result, became more creative and confident. While Loo averaged 11.1 seconds at the end of the official laps, the AI-controlled drones achieved an average of 13.9 seconds, reported NASA.
"Our autonomous drones can fly much faster. One day you might see them racing professionally," Reid said.
Camera-based localization and mapping technologies have a range of probable applications, according to Reid. These technologies can help drones to keep an eye on the warehouses or to lend a hand to the humans in search and rescue missions at disaster sites. Autonomous drones can also be used to assist robots in navigating a space station's corridors in the future.