Sony's AI-Powered Humanoid Robot Triumphs in Beijing Competition

An autonomous table tennis robot developed by Sony AI has successfully competed against and defeated top-ranked human players in official matches, according to Reuters. This system falls under the emerging field of "physical AI," where artificial intelligence is integrated into machines that operate in real-world settings.

The robot, named Ace, was engineered for competitive sports, demanding rapid decision-making and precise motor control. The project team reports that it merges high-speed perception with AI-driven control to execute shots under match conditions.

Ace participated in matches governed by International Table Tennis Federation rules and overseen by certified umpires. In documented trials from April 2025, the system won three out of five matches against elite players, losing two to professional-level opponents. Sony AI noted subsequent victories in matches held in December 2025 and early 2026 against professional players.

While table tennis robots have existed since the 1980s, earlier models could not compete at the level of advanced human players. "Unlike in computer games, where AI has already surpassed human experts, physical and real-time sports like table tennis present a significant ongoing challenge," stated Peter Dürr, director at Sony AI Zurich and the project lead.

Dürr highlighted that AI systems have excelled in fully simulated digital environments such as chess and video games.

He explained that the system was developed to explore how robots can operate with speed and accuracy in dynamic settings. The research was detailed in a study published in the journal Nature.

The sport poses technical challenges due to the ball's high speed and variability, including complex spins and changing trajectories, which require extremely fast sensing and coordinated movement. Ace's architecture incorporates nine synchronized cameras and three vision systems to track the ball's movement and spin. It processes visual data at a speed capable of capturing motion too fast for the human eye to perceive clearly. "This is fast enough to capture motion that would be a blur to the human eye," Dürr said.

The robotic platform uses eight joints to control the racket. Three manage positioning, two control orientation, and three regulate shot force and speed. This configuration was designed to meet the minimum mechanical requirements for competitive play.

Unlike many AI systems trained on human data, Ace was trained primarily in simulation. This method allowed it to develop its own unique strategies, leading to play patterns that differ from human opponents. Dürr noted the system "learns to play not from watching humans" but through self-training in simulated environments.

Professional player Mayuka Taira, who lost a match to the robot, found it difficult to predict because it displays no visible cues during play. Rui Takenaka, an elite player who both won and lost against Ace, observed that it handled complex spins effectively but was more predictable on simpler serves. Taira added that the robot's lack of emotional signals made it harder to anticipate its responses. "Because you can't read its reactions, it's impossible to sense what kind of shots it dislikes or struggles with," she said.

Dürr stated the system demonstrates a strong ability to read ball spin and react swiftly, with ongoing work focused on enhancing its adaptability during matches. The project team suggested that similar perception and control techniques could be applied to fields like manufacturing and service robotics.

Humanoid robots tested in long-distance race

At the 2026 Beijing E-Town Humanoid Robot Half Marathon, humanoid robots competed on a 21-kilometer course in Beijing. The event featured over 100 robots alongside approximately 12,000 human participants, who ran on separate tracks.

A robot named Lightning, developed by Honor, completed the race in 50 minutes and 26 seconds. This time was faster than Olympic runner Jacob Kiplimo's record of 57 minutes and 20 seconds set at the Lisbon Half Marathon in March. Lightning collided with a barricade during the race but continued to finish first. Honor robots also secured second and third place. Performance showed improvement from the previous year's event, where the fastest robot finished in two hours, 40 minutes, and 42 seconds. Organizers said the event aimed to test humanoid robots in large-scale, real-world conditions.

According to the Associated Press, another Honor robot completed the course in 48 minutes under remote control. However, race rules prioritized autonomous navigation, and Lightning was recognized as the official winner.

Honor engineers stated that technologies developed for the robot, including structural reliability and liquid-cooling systems, have potential applications in industrial scenarios.

See also: Cadence expands AI and robotic partnerships with Nvidia, Google Cloud

Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expo taking place in Amsterdam, California, and London. This comprehensive event is part of TechEx and is co-located with other leading technology events. Click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.