SHORT NEWS
University of Zurich teaches drones tricks
Thanks to a navigation algorithm, drones are learning to autonomously perform acrobatic maneuvers. Using simulations, the autonomous aircraft are trained to become faster, more agile, and more efficient.
According to a statement issued by the University of Zurich (UZH), the ability of quadrocopters to master maneuvers such as the Matty Flip or the Power Loop is “also beneficial during conventional search and rescue missions”.
A drone that is able to perform such acrobatic maneuvers can be much more efficient: “It can push itself to its physical limits, make full use of its agility and speed, and cover longer distances within its limited battery life.”
While during special competitions, professional human pilots try to outshine each other with drone acrobatics, in the case of the drone programmed at the UZH, the aircraft teaches itself tricks.
A certain amount of training is still necessary
The researchers developed the navigation algorithm that enables a drone – equipped only with on-board sensors and a processor – to independently perform various flight elements. In order to demonstrate the efficiency of their own algorithm, they flew maneuvers such as the Power Loop, the barrel roll and the Matty Flip, during which the drone is subjected to very high thrust and extreme angular acceleration.
A few hours of simulation training are all it takes to get the quadrocopter ready to fly without the need for further fine-tuning with real-life data. This is possible because the algorithm abstracts the learned inputs from the simulations and transfers them to the physical world.
Pizza delivery as an aerobatics show?
“With this navigation algorithm, we have reached another milestone on the path towards integrating autonomous drones into our everyday lives,” says Professor Davide Scaramuzza, Director of the Group for Robotics and Perception at the University of Zurich.
However, the researchers admit that human pilots still have the edge: “They can rapidly interpret unexpected situations and changes in the environment and adapt more quickly,” Davice Scaramuzza explains. Nevertheless, he is convinced that drones performing search and rescue missions or delivery services will benefit from being able to cover long distances quickly and efficiently.