Drone pilot

Are Humans More Proficient at Drone Piloting than AI?

Admin UAV Drones

According to a large number of people, Artificial Intelligence is the wave of the future. These individuals are of the mindset that one day, very soon, the need for the presence of an actual human being in the military, workforce, or to perform tasks even within our own homes may become obsolete. It is their contention that artificial intelligence is more effective, more productive and more precise. Recently, however, a demonstration conducted by two NASA researchers proved the dawn of the machines is not as imminent as some would have us believe, especially with respect to drone piloting and robotics.

Researchers selected drone racing as the platform from which to test the AI’s as a result of the fact that drone piloting requires dexterity and accurate maneuvering. In order to better understand the limitations of the AI’s capabilities, timed trials on a winding race track were executed against a world-class drone pilot Ken Loo. The results of drone pilot Loo versus the machines were surprising.

In October of this year, scientists at NASA’s Jet Propulsion Laboratory decided to wage a competition between an actual professional drone pilot and a robotic pilot. The demonstration was in an attempt to test two years of Google-funded research focused primarily on drone self-determination. The researchers designed and constructed three drones, appropriately nicknamed Batman, Joker, and Nightwing, and implemented algorithms that utilize Tango. Tango is a program developed by Google that makes use of computerized visualizations to aid in location services and navigation for smartphones and handheld devices.

The hope is that one day, the Tango platform, via its 3d mapping capabilities, may be used to help artificial intelligence systems, e.g. robotics, navigate the space station unassisted. In the meantime, researchers are testing the limits of its potential on drone and UAV technologies.

Each of the three drones had been developed according to racing specification with the capacity to achieve speeds up to 80 miles per hour in straight lines, however as a result of the winding track they were expected to race upon, acceleration of upward to roughly 40 miles per hour seemed a more realistic approach.

According to the research team’s task manager Rob Reid, their interest was in pitting their algorithms against human pilots who navigate more by ‘feel’. He said, “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 result was that at times, the robotic piloted drones moved so rapidly that motion blur caused disorientation. By the same token, although pilot Loo could perform impressive aerial maneuvers and could reach higher speeds, he became tired and was not able to perform for as longs as the AI, which does not experience exhaustion.

According to Loo, the track was the densest he has ever piloted. He stated,”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.”

Task manager Reid indicated that one day the population could see their robotic pilots racing drones professionally as they are capable of very fast flying. However, until that day, human drone pilots reign supreme and we’re certain they would relish the opportunity to test their skill against robotic aviators in the future.

Throughout the competition, it was evident that although the AI pilot demonstrated a more smooth drone flying experience, it wasn’t capable of matching the skill of drone pilot Ken Loo. Loo was faster on official laps with an average time of 11 seconds in comparisons to the AI’s 13.9.