Human pilot wins in race against AI-powered drones integrated with technologies from NASA and Google

The algorithms for controlling the drone were integrated with Google's Tango technology.

Researchers at NASA's Jet Propulsion Laboratory set up an obstacle course for drones in an empty warehouse. The obstacle course was to pit two years of AI drone development in collaboration with Google, against a world class drone pilot, Ken Loo. The drones used were called Joker, Batman and Nightwing, and were configured to fly at 129 kph in a straight line. However, while navigating the obstacle course, the drones could only reach a top speed of 64 kph.

Human pilot wins in race against AI-powered drones integrated with technologies from NASA and Google

Image: NASA

Loo was allowed a number of trial laps before the official timed laps. Initially, his timing matched that of the drones, but he became faster as his familiarity with the track improved. In the timed laps, Loo averaged a time of 11.1 seconds, while the AI-driven drones managed an average time of 13.9 seconds. The AI drones were fairly consistent in their runs, with less variation in the timings over several laps. Loo showed more variance in the timing and did not stick to a consistent path.

Rob Reid, the task manager of the project at JPL said, "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."

The AI drones were more cautious in their approach, and could not move too fast at times because the camera could not process the rapid movement. Loo had the luxury of managing the sudden burst of acceleration, but his movement around the path was less fluid. The drones could not depend on GPS for navigation and had to use cameras considering the small indoor location. The algorithms for controlling the drone were integrated with Google's Tango technology.

NASA has also contributed to the development of the Tango AR platform. One of the key findings was that while human drone pilots could tire after a few runs, the AI-driven drones had the advantage over several runs as they did not experience any fatigue. The researchers at JPL believe that in the future, the AI can catch up to humans. Reid said, "Our autonomous drones can fly much faster. One day you might see them racing professionally!"

Future applications for the drones include inventory checking in warehouses or providing assistance in search and rescue operations.

Find our entire collection of stories, in-depth analysis, live updates, videos & more on Chandrayaan 2 Moon Mission on our dedicated #Chandrayaan2TheMoon domain.

Loading...




also see

science