Researchers at ETH Zürich’s Robotic Systems Lab have taught quadrupedal robots to play badminton, showcasing a remarkable breakthrough in whole-body coordination. The video demonstrates how these legged manipulators use reinforcement learning to master the complex task of tracking and striking shuttlecocks while maintaining balance and mobility. What makes this achievement particularly impressive is the unified control policy that coordinates perception, locomotion, and precise arm movements simultaneously—a challenging triumvirate that has long stumped roboticists. The research proves that robots can now predict object trajectories, navigate dynamically, and interact with human players in fast-paced sports scenarios, opening new possibilities for robots in dynamic environments requiring dexterity and coordination.
