AI pilot defeats human champions in drone racing challenge

The University of Zurich, in collaboration with Intel, has unveiled its AI piloting system, "Swift," which recently defeated three world champion drone racers in a competition.

Swift's victory marks a significant advancement in AI's ability to navigate complex and dynamic environments.

The Swift AI system is the result of extensive research in AI and machine learning, building on previous iterations. Unlike traditional self-guided drones, Swift relies on real-time data from an onboard camera and an integrated inertial measurement unit to track its position, acceleration, and speed. This data is then processed by a deep neural network, which calculates the optimal trajectory for racing.

In a recent test, Swift achieved the fastest lap time, outperforming the human racers by half a second.

Davide Scaramuzza, head of the Robotics and Perception Group at the University of Zurich, highlighted the challenges of applying AI to physical sports, citing the unpredictability of real-world environments. The research team aims to further develop the Swift algorithm for applications in search and rescue operations, forest monitoring, space exploration, and film production.