The racing drones between human pilots and based on artificial intelligence ended with the victory of the first over the second, until the development of a new algorithm that allowed the autonomous drone not only win but have a more consistent flight.
An algorithm developed by researchers at the University of Zurich allowed a quadrotor, a drone with four rotors, was faster in autonomous flight than the human pilots it was competing against.
Drones, such as those used in rescue or order delivery, have in the autonomy of your battery one of the weak points, and that forces them to do the tasks assigned to them as quickly as possible.
And they may have to do it by going through a series of points of reference such as windows, rooms or specific locations to inspect, adopting the best trajectory and the correct acceleration or deceleration in each segment.
Drones evolve at the rate of artificial intelligence.
The new algorithm guides an autonomous drone, piloted by an AI system, on the fastest trajectory through a series of landmarks on a circuit.
For it, generates “optimal time trajectories that fully consider the limitations of drones […] rather than assigning sections of the flight path to specific waypoints, “as Davide Scaramuzza, director of the Robotics and Perception Group at the University of Zurich, explains.
The algorithm “just tells the drone to go through all the waypoints, but no how or when to do it“said Philipp Foehn, a doctoral student and first author of the study.
Smart drones can be useful for package delivery, inspection, search and rescue. Photo: AFP
Researchers measured the performance of the new algorithm against two human pilots in a race. The three participants were able to train in the circuit previously to be on equal terms.
To ensure a fair comparison, the human pilots had the opportunity to train on the circuit before the race. But the algorithm won: all his laps were faster than human laps and the performance was more consistent.
It was the first time that an autonomous quadrotor defeated human pilots in the race, as noted in the statement released to the press.
This is not surprising, because once the algorithm found the best trajectory, it can reproduce it faithfully many times, unlike human pilots, who obviously take different trajectories every lap.
The algorithm, the key
Before commercial applications, the algorithm will have to become less computationally demanding, as it now takes the computer up to an hour to calculate the optimal trajectory in time for the drone.
Also, at the moment, the drone depends on external cameras to calculate where it was anytime.
In future work, scientists want to use cameras on board. But the demonstration that an autonomous drone can, in principle, fly faster than human pilots is promising. “This algorithm can have enormous applications in drone package delivery, inspection, search and rescue, and more,” Scaramuzza completed.
With information from agencies.