Swift AI drone beats humans in drone race for the first time
Researchers from the University of Zurich and Intel have announced the creation of a unique autonomous unmanned system called Swift. It is powered by artificial intelligence and is capable of outperforming humans in FPV drone races, writes Ars Technica.
It is a sport in which participants try to guide high-speed drones through an obstacle course as quickly as possible. The pilots control the drones remotely, wearing a headset that transmits video from an onboard camera, giving them a first-person view.
Swift uses data collected in real time by the onboard camera and also measures the speed of movement. The AI processes the information to localize the drone in space and identify race gates on the track. Ultimately, this data helps to choose the most optimal course of action to complete the track as quickly as possible.
After training, the system competed with three world-class human pilots on a special track. For the first time, Swift managed to win several races and even recorded the fastest lap. However, the system demonstrated certain limitations in adaptability when the conditions differed from those on which it was trained.
The developers are convinced that Swift’s capabilities can be applied in the real world. For example, during space exploration or in the film industry, where high-speed drones can shoot action scenes. The technology can also be useful in search and rescue missions, where drones need to cover large areas quickly.