If you weren’t lucky enough to have a video game console as a kid, then you were probably stuck playing with one of the saddest toys ever devised: paddle ball. An elastic cord tethers a rubber ball to a small paddle, and the only goal is to keep bouncing the ball until you miss or give up from boredom. But while paddle ball isn’t fun, it is a great demonstration of coordination and dexterity. A robot called Cassie Cal from UC Berkeley has been taught to play paddle ball on its head.
The word “head” here might be misleading, as this robot doesn’t have any strong resemblance to a human — or any animal for that matter. It’s really just a pair of legs, and the paddle is attached at point just above the hips. Many robots are capable of tasks like juggling or keeping a ball bouncing on a paddle, but Cassie Cal’s unique body geometry and lack of arms adds an extra level of complexity to the task. Instead of simply moving an arm to keep the ball bouncing, Cassie Cal has to balance on two legs while moving her entire body to adjust the trajectory of the ball.
The roboticists working on this project, from the Hybrid Robotics Lab at UC Berkeley, only wanted to focus on that movement. They used a separate camera-based motion capture system to keep track of the ball, and that system provides Cassie Cal with the information she needs to hit the ball just right and keep it bouncing. The purpose of this research isn’t entertainment, but rather to develop robust movement algorithms. A server can easily walk through a crowded restaurant while balancing a tray full of drinks, but that would be difficult for a robot to replicate. This kind of research should help change that.