For the first time, a self-driving drone has beaten in speed drones guided by real pilots: it has done so thanks to a new algorithm capable of calculating the optimal path that guarantees the best performance in the shortest possible time.
The result was obtained by researchers from the University of Zurich led by the Italian robotics expert Davide Scaramuzza.
The study, published in the journal Science Robotics, could have numerous practical implications, from the field of rescue to that of transport.
To be useful, in fact, drones must be fast: due to the limited duration of the batteries, they must be able to carry out their mission in the shortest possible time, adopting the best trajectory with acceleration and braking in the right places. Until now, no autonomous flight system had managed to surpass the performance achieved by the pilots, but the drone developed by Scaramuzza did it, "beating the fastest lap achieved by two first-rate pilots", says the expert. The race took place on an experimental circuit under the eye of external cameras, used to follow the movements of the drone and to provide information on its position in real time to the algorithm. To ensure an equal challenge, the drivers were allowed to practice on the circuit before the race, but this was not enough: thealgorithm then beat them by conquering all the fastest laps and achieving more consistent performances.
Before bringing the drone to the market there are still some critical issues to be solved. Researchers are already working to simplify the algorithm (which currently requires almost an hour of calculations to determine the trajectory) and to install on-board cameras to monitor the flight.