New algorithm makes drones fly faster than world-class human racing pilots



[ad_1]

Drone flying through the smoke

A drone flying in the smoke to visualize the complex aerodynamic effects. Credit: Robotics and Perception Group, University of Zurich

To be useful, drones have to be fast. Due to their limited battery life, they must complete any task – looking for survivors at the site of a disaster, inspecting a building, delivering cargo – in the shortest possible time. And they may have to do this by going through a series of waypoints like windows, parts or specific locations to inspect, taking the best course and the right acceleration or deceleration at each segment.

Algorithm outperforms professional pilots

The best human drone pilots are very good at it and have so far always outperformed autonomous systems in drone racing. Now, a research group at the University of Zurich (UZH) have created an algorithm that can find the fastest path to guide a quadrotor – a four-propeller drone – through a series of waypoints on a circuit. “Our drone broke the fastest lap of two world-class human pilots on an experimental racetrack,” said Davide Scaramuzza, who leads the Robotics and Perception group at UZH and the Rescue Robotics Grand Challenge at NCCR Robotics. , which funded the research.

“The novelty of the algorithm is that it is the first to generate optimal trajectories over time which take full account of the limitations of drones”, explains Scaramuzza. Previous work was based on simplifications either of the quadrotor system or of the description of the flight path, and they were therefore suboptimal. “The key idea is, rather than assigning sections of the flight path to specific waypoints, that our algorithm simply tells the drone to go through all the waypoints, but not how or when to do it,” adds Philipp Foehn, doctoral student and first author. paper.

External cameras provide real-time position information

The researchers used the algorithm and two human pilots drove the same quadrotor on a racing circuit. They used external cameras to accurately capture the movement of the drones and – in the case of the autonomous drone – to give real-time information to the algorithm on where the drone was at all times. To ensure a fair comparison, the human pilots were given the opportunity to train on the circuit before the race. But the algorithm won: all of its tricks were faster than humans, and performance was more consistent. This is not surprising, because once the algorithm finds the best trajectory, it can faithfully reproduce it many times over, unlike human pilots.

Before commercial applications, the algorithm will have to become less computationally intensive, as it now takes up to an hour for the computer to calculate the optimal trajectory of the drone. Additionally, for now, the drone relies on external cameras to calculate where it was at all times. In future work, scientists want to use on-board cameras. But the demonstration that an autonomous drone can in principle fly faster than human pilots is promising. “This algorithm can have huge applications in parcel delivery with drones, inspection, search and rescue, etc.,” says Scaramuzza.

Reference: “Time-optimal planning for quadrotor waypoint flight” by Philipp Foehn, Angel Romero and Davide Scaramuzza, July 21, 2021, Scientific robotics.
DOI: 10.1126 / scirobotics.abh1221



[ad_2]

Source link