AI-powered badminton robot trained to play against humans

A Swiss-led team has developed an AI-driven legged robot that can play badminton against human opponents, demonstrating precise coordination and accuracy.

Developed at ETH Zurich and detailed in Science Robotics, the robot uses reinforcement learning, a type of AI that improves decision-making through repeated attempts.

The robot learns by trial and error to make better decisions.

Its AI combines vision, movement, and arm control, enabling the robot to track the shuttlecock and hit it accurately.

A perception noise model, trained on real-world camera data, ensures the robot maintains consistent performance.

"We introduced a perception noise model that maps the robot's motion to perception quality. And this allowed the reinforcement learning algorithm to automatically balance between the robot's agile motion and a reliable perception,'' said Yuntao Ma, a member of the team of researchers who have developed the badminton robot.

Ma believes this project showcases AI’s potential to advance legged robots in complex tasks, paving the way for advancements in autonomous, intelligent systems, including humanoids.

"The control algorithm also generalizes to other robot platforms such as humanoids and also other tasks such as search and rescue, and home services,'' he said. — Reuters