Drones

Human pilot holds off NASA's autonomous racing drones

Human pilot holds off NASA's autonomous racing drones
While the drone race was a fun way to test the progress of its AI-controlled drones, the JPL team is quite serious about the applications for its technology
While the drone race was a fun way to test the progress of its AI-controlled drones, the JPL team is quite serious about the applications for its technology
View 1 Image
While the drone race was a fun way to test the progress of its AI-controlled drones, the JPL team is quite serious about the applications for its technology
1/1
While the drone race was a fun way to test the progress of its AI-controlled drones, the JPL team is quite serious about the applications for its technology

Autonomous drones that can safely navigate obstacles have huge potential for everything from delivering goods to providing disaster relief, but they'll need to prove themselves first. NASA's Jet Propulsion Laboratory (JPL) recently pitted its autonomous racing drones against a human pilot, and although the AI competitors still have some ground still to make up, there were distinct advantages to the computer-driven approach.

The aircraft used for the showdown were developed as part of a research project at JPL, where its scientists are working on drones that use artificial intelligence, pre-loaded maps and built-in cameras to navigate their surroundings. This technology was originally developed with spacecraft in mind, which is what attracted the interest of Google who then funded the drone research spinoff.

The three drones used in the exercise, named Batman, Joker and Nightwing, were built to the same specifications as racing drones and were capable of hitting 80 mph (129 km/h). Professional drone racer Ken Loo was then invited to go head-to-head with them through a course set up in a JPL warehouse.

While Loo hit higher speeds overall, his more aggressive acceleration meant his movement was jerkier than the smooth-moving drones, which flew more cautiously. The complex algorithms sent the autonomous drones along pretty much the same path every lap, while Loo was able to be more creative and nimble, even pulling off corkscrew maneuvers.

This meant Loo averaged 11.1 seconds per lap, while the AI drones averaged 13.9 seconds. But it wasn't simply a question of speed. The JPL team points out that its drones flew more consistently overall and weren't troubled by fatigue, unlike Loo.

"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."

While the drone race was a fun way to test their progress, the JPL team is quite serious about the applications for its technology. These kinds of camera-based navigation technologies could be used in areas where GPS isn't available, such as in drones that check inventories in warehouses or assist in disaster operations, and even in robots trying to navigate space stations.

See the drones in action in the video below.

Source: NASA

Drone Race: Human vs. Machine

No comments
0 comments
There are no comments. Be the first!