Table tennis-playing robot on track to becoming world champion

Ace, an autonomous robot powered by AI, cutting-edge sensors and an extremely dexterous arm with eight joints, played table tennis according to competition rules and beat elite human competitors. The robot is the first machine to excel in this sport.

It was the game of brain chess that was first disrupted by computers, but Ace’s success suggests that physical sports may be about to have their “Deep Blue” moment – ​​the day in 1997 when a machine of that name beat world chess champion Garry Kasparov.

“Games have long served as a benchmark for AI, including chess for Deep Blue, but also other games with more recent advances, like [the Go-playing AI] AlphaGo,” says Peter Dürr of Sony AI, Zurich Switzerland, who led the team that built Ace.

But he says those earlier stages of AI took place online. Ace represents a significant step forward as he has faced real-world professional table tennis champions and held his own.

“Ace offers something that has simply never been captured before: a robot and a human in a real sporting competition,” says Dürr.

Ace has three main advances in autonomous robotics, he says. First, it uses “event-based sensors,” meaning the robot focuses on certain regions of the images captured by its cameras – those indicating changes in movement or brightness, which are key to tracking the trajectory of the table tennis ball.

Next, the robot’s table tennis skills are developed using “model-free reinforcement learning”, which means, according to Dürr, that the robot “learns through the experience of simulation rather than adopting a model of how table tennis should be played”. This process was similar to when the robot was playing a computer game of table tennis, and the robot had accumulated several thousand hours of training in the process.

And finally, the team deployed high-speed robotic hardware that allows Ace to play with “human-like agility,” Dürr says. In some ways, he’s even more agile than a human, because athletes need about 230 milliseconds to react, he says, while Ace’s total latency is only about 20 milliseconds.

Currently, the robot looks like a robot from a factory and relies on a network of cameras and sensors surrounding the table tennis arena. But as technology advances, researchers expect that Ace will eventually incarnate in humanoid form.

For matches played in a study published today, the rules of Japan’s professional table tennis league applied as Ace faced five elite but non-professional players, each of whom had competed for at least a decade and trained 20 hours a week. The robot also hired two professionals.

Ace has only lost two of his five matches against elite players, but his two matches against professional players. He did, however, achieve a victory in a match during one of the professional matches.

Another advantage Ace has over humans is that he gives no indication of his next move. On the other hand, he does not have the ability to read human body language signs.

“Some athletes involved in our experiments said that they usually watch their opponent’s face, which is not the case for Ace,” says Dürr.

Others were surprised by Ace’s ability to read their serve rotation, despite their attempts to hide it with different moves. The robot also baffled its inventors, especially when it was able to hit balls that bounced off the net, which was not a skill it had trained for. It’s a skill that “has just emerged,” Dürr explains.

Over the past year, since the study ended, the team has continued to improve Ace’s abilities.

In December 2025, Ace defeated a professional player for the first time, and in March 2026, Ace won matches against three other professional players: a female professional, Miyuu Kihara, ranked in the top 25 in the world table tennis rankings, as well as two male professionals, Tonin Ryuzaki and Fumiya Igarashi.

“With further improvements it should be possible to surpass even the world champion,” says Dürr.

And the improvements go both ways, he says.

“Former Olympian Kinjiro Nakamura noted that before watching Ace he thought a certain shot was impossible, but after seeing it he thinks human athletes could replicate the technique.”