Sony AI's table-tennis robot, named Ace, has made history as the first autonomous machine to compete against and occasionally defeat top-level human players in the fast-paced sport of ping-pong. Developed by the Japanese company's research division, Ace integrates high-speed perception, deep reinforcement learning, and agile robotics to handle the rapid decisions and precise execution required at professional speeds. According to project leader Peter Dürr, director of Sony AI in Zürich, this marks a major breakthrough, demonstrating that robots can match or exceed human reaction times and decision-making in real-world, interactive physical tasks.
Ace adheres to the official rules of the International Table Tennis Federation (ITTF) and has been tested against elite competitors. In matches against five elite amateur players—who practice an average of 20 hours a week and have over a decade of experience—Ace played best-of-three sets and secured three match wins out of 13 total games, winning seven games overall. It faced two professional Japanese league players, Minami Ando and Kakeru Sone, in best-of-five matches, winning one game out of seven but ultimately losing both encounters. As reported by Reuters via Slashdot, Ace's ability to return balls at speeds up to 19.6 meters per second allowed it to engage in intense rallies and execute competitive serves.
What sets Ace apart from earlier ping-pong robots, like Omron's FORPHEUS that challenged amateurs at CES 2017, is its elite-level performance against pros, as highlighted by The Verge. The robot excels not through brute force but superior control, successfully returning 75 percent of spinning balls across a wide range of spins—often surprising human opponents with precise maneuvers and direct points on serves. Sony AI's research page details how event-based sensing and real-time AI control enable this superhuman perception, bridging the gap from virtual simulations to physical competition.
This achievement underscores the growing capabilities of physical AI agents in complex, dynamic environments, with implications for robotics in sports, manufacturing, and beyond. Elite players and professionals are directly affected, as Ace pushes the boundaries of human-machine interaction in a sport known for its skill intensity. While Ace still trails top pros in overall matches, its wins signal rapid progress; future iterations could refine techniques against professional spins and strategies, potentially reshaping training tools or competitive exhibitions. Researchers emphasize that these results open doors to autonomous systems outperforming humans in high-stakes, real-time scenarios.