Sony AI's Project Ace Becomes First Robot to Beat Pro Table Tennis Players, Lands Nature Cover
A Sony research robot combining event-based vision and deep reinforcement learning has defeated elite and professional table tennis players in real matches, marking the first time a machine has reached human expert level in a fast, physical competitive sport.
Sony AI has published Project Ace on the cover of Nature, claiming the first autonomous robot to play competitively against elite and professional table tennis players in the real world. The paper, titled "Outplaying elite table tennis players with an autonomous robot," represents a long-standing benchmark in robotics: human expert-level performance in a fast, physical, adversarial sport — not in simulation, not at slow speeds, and not against amateurs.
The system fuses three ingredients that have been advancing in parallel for years and rarely converge in one platform. Event-based image sensors from Sony Semiconductor Solutions track the ball in 3D at 200 Hz with roughly 3.0 mm of error and 10.2 ms of latency. A deep reinforcement learning policy, trained largely through self-play and simulation, drives shot selection and placement. And a highly agile robotic arm executes returns with end-to-end latency around 20.2 milliseconds — more than ten times faster than the ~230 ms reaction time of an elite human player.
The match results back the claim. In December trials against four new opponents, Ace beat both elite-level players and one of two professionals, losing only to the second pro. In a follow-up round in March 2026 against three new professionals, Ace defeated each of them at least once. That record places the system in the same competitive band as serious human players, not just well-rehearsed amateurs — a threshold that vision-based table tennis robots have approached for years without crossing.
The bigger story is what Project Ace says about physical AI. Most frontier AI work in 2026 lives in language and code; embodied systems that perceive, decide, and act in the physical world at human reaction speeds remain rare. Sony argues that the same stack — event sensing, RL-trained policies, and low-latency hardware — generalizes well beyond sports and points toward industrial robots, household assistants, and autonomous vehicles that can finally close the loop fast enough to operate alongside people.
