NASA’s AI drone just barely loses to human racer

NASA’s latest AI drone can navigate an area with ease. In fact, it can almost, but not quite, keep up with a world-class racer.

Drones have (no pun intended) truly taken off in recent years, quickly becoming commonplace. In fact, they’re so common that professional drone racing has become a sport. Who better, then, than professional drone racer Ken Loo,
to take on NASA’s latest AI drone in a speed obstacle course? Well, if the aim was to spare the programmers’ feelings, then perhaps somebody less talented would have been better.

The product of two years of research by NASA’s Jet Propulsion Laboratory, the potential uses for a AI drones are numerous. With applications for wildlife study and research, search and rescue and, with modification, exploration on other planets, it’s easy to see the appeal. At the moment, these drones are really only capable of flying along a set path, rather than exploring autonomously. However, for this particular race, this may have made them all the more of a challenge.

Ken Loo, who races under the name Flying Bear, was the human chosen to challenge NASA's AI drone racers.
Ken Loo, who races under the name Flying Bear, was the human chosen to challenge NASA’s AI drone racers.

The challenge.

An indoor obstacle course provided the scene for Ken Loo to race against the three AI drones, names Batman, Joker and Nightwing. These racing drones are reportedly capable of reaching 80mph, though the cramped course restricted speeds to about 40mph. At first, Loo was unable to keep up with the AI drones, but once he got used to the twists and turns of the course, things changed. With practice and more understanding, Loo was able to finish well ahead of the computerised racers. While the AI drones followed the same path each and every time, Loo was better able to adapt. The human racer took more risks and challenging turns, giving him the deciding edge overall. That said, the racer was able to spot a weakness that might have let the computers win out in the long term. 

“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.”

Of course, the computerised flyers don’t suffer from the same vulnerability. This could make them ideal for long term navigation or surveillance, where humans just can’t cut it. We’ve all seen technology advance with incredible speed over the years, so perhaps Loo’s victory is only temporary. The true test for AI drones will be out in the real world. Seeing how these drones can navigate a random, changing, perhaps even hostile, environment will be the true testament to their abilities.

No Comments Yet

Leave a Reply

Your email address will not be published.