There are many reasons for drones to be quick. The professional drone racing circuit aside, speed bodes well when you are searching for survivors on a disaster site, or delivering cargo, or even inspecting critical infrastructure. But how do you get something done in the shortest possible time with limited battery life when you have to navigate through obstacles, changing speeds, and altitude? You use an algorithm.
Researchers at the University of Zurich (UZH) have developed an algorithm that can find the quickest trajectory to guide an autonomous drone through a series of waypoints on a circuit.
This algorithm is not just good, it’s extremely good. As Davide Scaramuzza, who heads the Robotics and Perception Group at UZH, explains:
Our drone beat the fastest lap of two world-class human pilots on an experimental race track.
Racing against an AI-driven drone
Here’s how that race went down: The researchers had the algorithm and two human pilots fly the same drone through a race circuit. There were external cameras on the scene. These cameras severed two purposes:
- They captured the precise motion of the drones, and
- In the case of the autonomous drone, they provided real-time information to the algorithm on where the drone was at any moment
To be fair, the researchers did give the human drone racing pilots time to train on the circuit before the race. But the algorithm won. Not only were all its laps faster than the human ones but its performance was also more consistent. And that’s because, unlike human pilots, once the algorithm has found the best trajectory, it can just mimic it flawlessly as many times as needed. Tells Scaramuzza:
The novelty of the algorithm is that it is the first to generate time-optimal trajectories that fully consider the drone’s limitations.
Philipp Foehn, a Ph.D. student and first author of the paper, adds:
The key idea is, rather than assigning sections of the flight path to specific waypoints, that our algorithm just tells the drone to pass through all waypoints, but not how or when to do that.
Ready for primetime use?
Now, the algorithm may have proven a success, it’s not ready for commercial applications yet. And that’s because its computational demands are huge. The computer needs up to an hour to calculate the quickest trajectory for the drone. Also, right now, the drone relies on external cameras to figure out where it is at any given moment. In the future, scientists want to use onboard cameras for spatial cognition.
So, while they continue their work on the novel algorithm, you can check out their real-world agile flight footage to learn more about the methods applied: