Swift AI, built by a team at the University of Zurich, brought human world champions in drone racing down to earth with a new speed record.
In case you didn’t know drone racing world championships were a thing, a team of Swiss researchers have some interesting news for you: AI is the sport’s new world champion.
An AI system called Swift, designed by scientists at the University of Zurich, has defeated three world champions in drone racing and clocked the fastest lap on record. This makes it the latest sport in which AI has outperformed humans, after chess, Go and StarCraft.
Swift AI won 15 of 25 races against the current world champions using a technique called deep reinforcement learning, detailed in a study published in Nature this week.
“Our result marks the first time that a robot powered by AI has beaten a human champion in a real physical sport designed for and by humans,” Elia Kaufmann, a researcher who helped to develop Swift, told The Guardian.
“This is the start of something that could change the whole world. On the flip side, I’m a racer, I don’t want anything to be faster than me,” added Thomas Bitmatta, one of the three champions.
The university’s robotics and perception group have published a video explaining the tech behind Swift AI and how it was able to beat expert-level human players in drone racing.
In it, Kaufmann and his team explain how Swift AI uses a technique called visual-inertial odometry (VIO) to estimate its own position, velocity and orientation. It also helps that, unlike human players, AI never gets tired.
“This work represents a milestone for mobile robotics and machine intelligence, which may inspire the deployment of hybrid learning-based solutions in other physical systems,” the team wrote in the paper.
However, as AI and drones are being increasingly deployed in a wide range of human activity, experts such as Dr Elliot Winter, a senior lecturer at Newcastle Law School who spoke to The Guardian, warn that governments should err on the side of caution when using AI in the military.
“We must be careful not to assume that advancements such as these can easily be transplanted into a military context for use in military drones or autonomous weapons systems which are involved in critical processes such as target selection,” he said.
10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.