Smash and Serve: Four-Legged Robot Takes on Humans in Badminton Showdown!

A dedicated group of engineers at the Robotic Systems Lab, ETH Zurich, Switzerland, has developed, constructed, and evaluated a four-legged robot that can play badminton alongside human competitors.

In their research, published in the journal Science Robotics, the team utilized a reinforcement learning-based controller that enables the robot to track, anticipate, and react to the shuttlecock’s movements during play, showcasing the potential of multi-legged robots in fast-paced sporting environments.

Badminton, akin to tennis, differs primarily in the use of a shuttlecock instead of a tennis ball. The objective remains the same: to hit the shuttlecock over a net positioned at midcourt to an opponent waiting on the other side.

Playing badminton demands agile footwork for quick positioning and skillful arm and hand coordination to hit the shuttlecock accurately and direct it over the net to a specific target. Equipping a robot with these skills necessitated several modifications.

The primary modification involved giving the robot four legs instead of the conventional two utilized by humans. This design enhances the robot’s stability and movement agility.

To enable these capabilities, the researchers equipped the robot with a stereo camera and a dynamic arm. They also implemented a reinforcement-learning-based controller that allows the robot to adjust its position and respond effectively to the shuttlecock.

Additionally, the team employed a “perception noise model” to align real-time data from the camera with the information stored in the database created during the training phase. This innovation allowed the robot to perform human-like actions in badminton, such as following through on strikes and tilting its base to maintain sight of the shuttlecock.

Testing confirmed the effectiveness of their training; the robot, known as ANYmal-D, was able to navigate the court proficiently and sustain rallies with human players for up to 10 shots.

© 2025 Science X Network