Artificial intelligence can beat us at chess, Go, the game show Jeopardy, and a growing number of different occupations in the workplace. But how does it fare against human pilots when it comes to challenging them at the exhilarating new sport of drone racing? Researchers at NASA’s Jet Propulsion Laboratory in Pasadena, California, recently decided to find out — by pitting a professional human drone pilot against racing drones controlled by cutting-edge A.I. machine intelligence.
Held on October 12, the Google-funded race involved a timed trial between A.I. drones named Batman, Joker, and Nightwing, powered by Google’s Tango technology. Against them — serving as the representative of non-Google, non-NASA humanity — was world-class drone pilot Ken Loo.
“We pitted our algorithms against a human, who flies a lot more by feel,” Rob Reid, of the Jet Propulsion Lab, said in a statement. “You can actually see that the A.I. flies the drone smoothly around the course, whereas human pilots tend to accelerate aggressively, so their path is jerkier.”
Loo described the obstacle course as, “definitely the densest track I’ve ever flown.” He was able to reach the higher speeds during the race and pulled off more impressive acrobatics. Ultimately, he averaged 11.1 seconds for his official laps. However, he also admitted to getting “mentally fatigued” by the demanding course and flew less consistently than the computer-controlled drones, which averaged 13.9 seconds.
In other words, humanity came out of the competition as messier, but more creative, whereas the A.I.-powered bot exhibited machine-like precision and consistency.
The results of the October 22 race put humans in the winning position for now, although the A.I. drones still performed impressively well. In fact, NASA seems unwilling to cede victory to us fleshy humans for long, since Reid insists that, “Our autonomous drones can fly much faster. One day you might see them racing professionally!”
Looking forward, NASA hopes that the camera-based localization and mapping technologies involved in the research could be used for a variety of applications — such as checking inventory in warehouses, performing search and rescue missions at disaster sites, or even helping robots navigate through space stations.
Published at Wed, 22 Nov 2017 22:19:01 +0000