MIT Teaches Mini Cheetah Robots to Play Soccer Goalie Using AI
Researchers utilize a reinforcement learning framework to train MIT's Mini Cheetah robots to block soccer balls in under a second. The same underlying technology shows promise for future offensive robotic soccer skills.
Researchers apply a model-free reinforcement learning framework to teach MIT's Mini Cheetah robot how to play soccer goalie. This complex challenge requires the quadruped to combine highly dynamic movement with precise, fast ball manipulation to block incoming shots in under a second.
The robotic system relies on three primary moves—sidestep, dive, and jump—to intercept the ball while calculating its trajectory on the fly. To prove the effectiveness of this training method, the team pits the AI-powered goalie against both human players and a second Mini Cheetah robot on the soccer field.
While the current research focuses entirely on defensive goalkeeping skills, the authors note that this hierarchical framework extends easily to other scenarios. Scientists plan to adapt the same underlying technology to teach the robots multi-skill offensive maneuvers, such as kicking soccer balls down the field.