Sony says its table-tennis robot, Ace, has reached expert human performance and occasionally defeats professional players—an advance the company calls a milestone for AI and robotics. Detailed in Nature, the system relies on reinforcement learning, nine cameras around the court to track ball spin via its logo, and an eight-joint arm to execute shots at humanlike speed and reach. Researchers staged matches on an Olympic-sized court in Tokyo, with Japanese Table Tennis Association umpires, and report Ace beat most of four high-skill opponents in December.
Sony AI president Michael Spranger said the goal was comparability, not superhuman power—winning through tactics and decision-making rather than raw speed. He argues the approach signals broader industrial potential where robots must adapt in fast-changing settings, though he acknowledged implications for defense.
Veteran roboticist John Billingsley praised the achievement while noting the robot’s extensive sensing gives it an edge over two-eyed humans. Sony frames Ace as the first robot to attain expert-level play in a widely played physical sport, part of what Spranger calls a “ChatGPT moment for robotics” as AI-driven methods push from simulation to real-world agility.
Related article:
