Sony AI desk tennis robotic outplays elite human gamers

Ace rotates its paddle because it prepares to return the ball again to its human opponent, Yamato Kawamata, throughout a match in December 2025. Credit score: Sony AI.

In an article printed at this time in Nature, Sony AI introduce Ace, the primary robotic to beat elite human gamers in aggressive bodily sport.

Though AI techniques have proven superior efficiency in digital domains and board video games (similar to advanced video video games, chess and Go), translating this to bodily efficiency has remained a big problem. Such a feat requires notion, planning, and management to work in a high-speed area on the size of milliseconds. Desk tennis is a demanding and complicated real-world check for robotics, requiring speedy decision-making, exact bodily execution, and steady adaptation to an unpredictable opponent. The ball’s excessive pace, spin, and complicated trajectories are central to aggressive play.

Director of Sony AI in Zürich, and challenge lead for Ace, Peter Dürr stated “this analysis has proven that an autonomous robotic can, in actual fact, win at a aggressive sport, matching or exceeding the response time and determination making of people in a bodily area. Desk tennis is a sport of monumental complexity that requires split-second selections in addition to pace and energy. This analysis breakthrough highlights the potential of bodily AI brokers to carry out real-time interactive duties, and represents a big step towards creating robots with broader functions in quick, exact, and real-time human interactions.”

A whole view of desk tennis robotic, Ace, together with arm and observe. Credit score: Sony AI.

What new parts does Ace incorporate?

Ace combines event-based imaginative and prescient sensors and a management system based mostly on model-free reinforcement studying, in addition to state-of-the-art high-speed robotic {hardware}. Ace was designed with three novel parts:

• A excessive pace notion system composed of 9 energetic pixel sensor cameras to find out the ball’s exact 3D place, mixed with three gaze management techniques that use event-based imaginative and prescient sensor cameras, pan/tilt mirrors, and telephoto tunable lens to measure the ball’s angular velocity and spin in actual time.
• A novel management system based mostly on model-free reinforcement studying to allow speedy adaptation and decision-making with out reliance on pre-programmed fashions.
• Excessive-speed robotic {hardware} able to executing exact, high-speed management for agile bodily interplay.

Members of the Ace analysis crew and desk tennis officers pose with the robotic and its human opponent, Mayuka Taira, following an official match in December 2025.

From Determine 4 within the Nature manuscript “Outplaying elite desk tennis gamers with an autonomous robotic” this movie reveals the robotic making a cut up part change to its trajectory when the ball hits the web. Credit score: Sony AI and Nature.

Testing Ace towards elite gamers

For the outcomes reported within the Nature publication, Ace was evaluated in matches towards 5 elite gamers and two skilled desk tennis gamers, beneath Worldwide Desk Tennis Federation (ITTF) laws. Ace achieved three victories in 5 matches towards the elite gamers, together with aggressive performances within the remaining matches.

There have been some fascinating outcomes from the evaluations, together with the truth that Ace was in a position to return a variety of spins, persistently reaching over 75% return fee as much as spins of 450 rad/s. The management techniques behind Ace additionally allowed for fast response to uncommon pictures, similar to balls bouncing off the web. This habits illustrates the power of the strategy to generalize to conditions which can be each uncommon and exhausting to mannequin in simulation.

Following submission of the Nature manuscript, the crew carried out further aggressive matches in December 2025 and March 2026, beating skilled gamers within the course of. In contrast with earlier evaluations, Ace demonstrated increased shot speeds, extra aggressive placement nearer to the desk edge, and faster-paced rallies, reflecting continued efficiency beneficial properties beneath aggressive circumstances.

Discover out extra concerning the challenge on this video from Sony AI.