A fully automatic table tennis robot named "Ace" recently won several official matches against top human players in Tokyo. It is regarded as a landmark event that artificial intelligence and robotics technology have reached a new milestone in the field of sports.
"Ace" was developed by the artificial intelligence research department of Japan's Sony Corporation. It is the first robot system to reach expert level in highly competitive physical sports events that require extremely high speed and precision. The project leader said that Ace relies on high-speed visual perception, artificial intelligence control algorithms and advanced robotic hardware platforms to make extremely fast and precise technical choices and shot executions in table tennis matches.
Various table tennis robots have appeared in the past since 1983, but they have always been difficult to compete with high-level human players. The emergence of Ace has changed this situation: it has played against elite and professional players in matches that comply with the official rules of the International Table Tennis Federation and are officiated by certified referees, and has won some of the duels. Project leader Peter Dell, head of Sony AI Zurich, said that unlike "pure digital" AI that has already surpassed human experts in chess and video games, physical sports such as table tennis that require close-range and high-speed confrontation are still regarded as a major problem in the field of AI and robotics.
Del pointed out that the goal of the project is not only to let robots compete with humans at the table tennis table, but more importantly, to explore how robots can achieve a human-like "perception-decision-action" closed loop in a dynamic environment and complete prediction and control in a very short time. He emphasized that Ace's success in visual perception and learning-based control algorithms means that similar technologies are expected to be expanded to other scenarios that require high-speed real-time control and interaction with humans, such as manufacturing, service robots, and a variety of application fields covering sports, entertainment, and even safety-critical physical systems. Relevant research results have been published in the journal "Nature" published on the 22nd.
The research team disclosed in the paper that Ace won 3 out of 5 games against elite players in April 2025 and lost in two games against professional players. Sony AI added that in December 2025 and March this year, Ace has achieved victory over professional players, and its performance has continued to improve. At the same time, companies around the world continue to make breakthroughs in the field of robots. For example, in the half marathon held in Beijing this week, robot runners outperformed human runners.
Different from computing in a virtual environment in chess or video games, table tennis competitions require the system to make instant decisions while completing precise hits on high-speed flying balls, and continuously adapting to changes that are difficult to predict by the opponent. Table tennis balls fly at high speeds with strong rotation and complex arcs, pushing both humans and robots to the limits of perception, prediction and motion control capabilities. To this end, Ace's architecture integrates 9 synchronized cameras and 3 sets of vision systems to track high-speed rotating ping pong balls in real time with extremely high precision. Dell said the system is responsive enough to capture subtle changes in motion that appear almost as a "blur" to the human eye.
In terms of body design, Ace uses a customized robot platform with 8 joints. Del explained that this is the "minimum configuration" for performing competitive competitive shots: 3 joints control the racket position, 2 joints are responsible for the racket posture, and the other 3 joints are used to adjust the speed and power of the bat. This design allows Ace to complete a variety of high-quality shots, including quick pulls, line changes, and responses to different spins.
Judging from the feedback from human opponents, Ace's "style" on the court is also very oppressive. Professional player Mayuka Taira, who lost a match with Ace last December, said the robot was "very unpredictable and completely emotionless." She admitted that because it is impossible to judge preferences and weaknesses from their expressions or body reactions, it is almost impossible to "read the opponent" to find out the ball paths that they are not good at dealing with, which greatly increases the difficulty of the game.
Elite player Rui Takenaka, who has played against Ace many times and has won and lost, believes that the robot's "ability to identify" different serve qualities is impressive. He said that when he served with complex spin, Ace would also fight back with complex spin, which made him very difficult in the subsequent stalemate; and when he switched to a weakly rotating "no-spin ball" (commonly known as "floating ball" or "edge without spin") to serve, Ace's return was relatively simple, thus creating opportunities for him to attack on the third board. This was regarded by him as one of the key factors in defeating Ace.
Although Ace has demonstrated "superhuman" capabilities in multiple indicators, Del believes that this robot system still has a lot of room for improvement. He pointed out that Ace has reached a level beyond humans in reading the spin and reaction speed of incoming balls, and because it does not learn by watching human games, but mainly through simulated self-confrontation training, it often shows a completely different response mode from humans in the game, and even creates situations that even experienced players are not expecting. However, in contrast, professional athletes are extremely talented at adapting to their opponents and finding weaknesses, which is one of the areas the research team is focusing on.