Table tennis robot defeats some of world’s best players – why this has major implications for robotics

A table tennis robot called Ace has outperformed elite players in recent evaluations. Developed by Sony AI, Ace won three out of five matches against competitive athletes with over ten years' experience. The robot lost to professional Japanese league players but won a game against one of them. Ace's performance is a significant step toward AI systems that can operate in fast, uncertain environments. Ace uses three 'event-based' vision sensors and nine high-speed cameras to track the ball and opponent. It employs deep reinforcement learning, trained in simulation over millions of virtual rallies. The system can estimate spin in real time, a longstanding challenge for robotic systems.