Researchers at the University of Zürich have developed a new algorithm to control autonomous flying of quadrotor drones. The new algorithm was put to test against a pair of human drone racing pilots and managed to beat them in a race. The algorithm calculates time-optimal trajectories that fully consider the drone’s limitations.

Drones have many potential applications in the real world, such as looking for survivors in disaster-affected areas and delivering goods. Drones have to be fast because they have a limited battery life to finish the assigned task.

Beating humans at their game

Needing to be fast also means finishing the task as quickly as possible, and many do traverse several waypoints such as doors and windows to inspect specific locations. The new algorithm enables them to adopt the best trajectory and the right speed to reach each place.

Usually, expert human drone pilots can beat autonomous systems, but that doesn’t seem to be the case anymore. The algorithm was able to guide the drone through a series of waypoints on the circuit. The algorithm managed to better the fastest lap of two world-class human pilots on an experimental track.

The algorithm is the first to generate time-optimal trajectories that fully take the limitations of the drone into consideration. The idea is here is to tell the drone to pass through all waypoints but not how and when to do it.

Many potential applications

For the research, the researchers had the algorithm and two human pilots fly the same drone around the circuit. External cameras were used to capture the motion of the drone and for it to provide real-time information to the algorithm of the whereabouts of the drone. Human pilots were allowed to take practice laps before the race.

Despite the practice, the algorithm proved to be faster, with all laps recording faster times than the pilots, and its performance was more consistent. Once the algorithm determines the best trajectory, it reproduces that trajectory many times while humans are unable to replicate the consistency. Researchers say their algorithm could have many used cases such as package delivery, inspection, search, and rescue operations.