Sony’s AI Table Tennis Robot Beats Elite Players, Marks First Expert‑Level Sport Performance

Context Sony built a paddle‑wielding robotic arm, called Ace, inside a full‑size table‑tennis arena at its Tokyo headquarters. The machine uses nine high‑speed cameras and eight articulated joints to track the ball and move the paddle. Researchers trained it with reinforcement learning, a method where the system improves by trial and error rather than being programmed with fixed rules.

Key Facts The robot faced professional athletes who train at least 20 hours a week and obeyed official table‑tennis regulations. In multiple matches, Ace matched or outperformed its human opponents, a milestone described by Sony as the first time a robot has reached expert level in a commonly played competitive sport. Peter Dürr, a Sony AI researcher, explained that “there’s no way to program a robot by hand to play table tennis. You have to learn how to play from experience.”

Speed remains a core challenge for robots in unpredictable settings, according to Michael Spranger, president of Sony AI. He noted that most fast robots operate in fixed‑trajectory factories, whereas Ace had to react to a constantly changing ball trajectory. The team calibrated the robot’s physical abilities to be comparable to a skilled human, avoiding a simple “shoot‑the‑ball‑faster” advantage. Instead, Ace won by making tactical decisions and adapting its strokes in real time.

What It Means Ace demonstrates that AI can acquire complex motor skills through experience, not just data‑driven text generation. The breakthrough suggests future robots could operate safely and effectively in dynamic environments such as disaster zones, manufacturing lines that require on‑the‑fly adjustments, or even assistive devices that need to anticipate human movement. The achievement also raises questions about how AI‑driven sports training might evolve, with machines offering real‑time strategic feedback.

Watch for further tests of Ace in varied sports and its integration into broader robotics platforms that demand rapid, adaptive decision‑making.