- A robot just beat elite table tennis players
- Sony AI’s Project Ace is effective at competing against unpredictable human players
- Success in this area could mean it will be easier to use AI to train future robots to deal with the real world
In competitive table tennis, the ball can travel at speeds of up to 70 mph and go anywhere. Sure, there is some predictability based on hitting, spinning, and how the ball hits the table, but there are also endless possibilities that a robot now seems to have mastered.
Sony AI’s Project Ace is the first robot to beat multiple elite-level table tennis players in an International Table Tennis Federation-style arena and under the watchful eyes of licensed referees.
In a new Nature article, Outperform elite table tennis players with an autonomous robotSony AI scientists describe their work and how they built and used AI to train a robot, “Project Ace,” not just to play table tennis, but to do it at a professional level.
Article continues below
“Ace achieved three victories in five matches against elite players, as well as competitive performances in the remaining matches. These results demonstrate the potential of physical AI agents to outperform human experts in real-time interactive tasks,” the scientists wrote.
Project Ace is a clever combination of “high-speed perception,” a reinforcement learning-based control system (rewarding good behavior), and “high-speed robotic hardware.”
No feet, but a nasty backhand
Ace doesn’t look like any human ping pong player you’ve ever seen. Instead, it glides in four directions on a custom track system, while its trunk rotates 360 degrees and the fully articulated arm and wrist adjust on the fly to serve and return the ball. You may have seen robots playing table tennis before (I remember seeing one at CES 2026), but not like this. The speed alone is astonishing.
Yet, it is the AI-based reinforcement learning and training simulation that makes Project Ace special and successful. In training, he was able to carry out all kinds of gaming scenarios. He even trained against a virtualized version of himself. But it is “model-free” reinforcement learning that allows, at least in part, Project Ace to adapt to unpredictable, elite human competitors.
Equipped with onboard sensors and an array of nine cameras positioned around the robot, Project Ace can see things that most human competitors, even elite players, might miss. The rotation of the ball, for example, is decisive in knowing where the ball will go next.
As the researchers explain in the project video, perception is one of the key innovations: “It is therefore the only system in the world capable of measuring the rotation of an unchanged table tennis ball at this speed. »
The secret sauce here may lie in a technique called “privileged criticism,” which Sony’s AI developers used in training simulations. The privileged reviewer accessed perfectly tailored information, combined with live sensor data. You could say that it is about comparing what should happen with what is happening. This learning allows the robot to prepare for the unexpected.
There’s a moment in the video where you can see this at work. The elite human player hits a ball that catches the net, sending it in a different, perhaps unanticipated, direction. Project Ace clearly already had a comeback planned, but he managed to adapt to the new ball trajectory and stage a comeback. Everything happens in milliseconds, and one could argue that a human player wouldn’t have been able to make the same rapid adjustment.
“It completely blew my mind,” Peter Dürr, Sony’s director of AI and chief engineer, said in a statement about Project Ace.
Like Sony AI’s previous project: teaching AI how to beat human-level expert players in a Gran Turismo simulation, Project Ace isn’t about beating professional players and advancing to the Olympic level of tablet tennis players. It’s about helping robots operate in an unpredictable world.
Most people who watch humanoid robots operating in home environments comment on their speed, or lack thereof. Robots move deliberately for safety reasons and to deal with unforeseen events. Project Ace, however, proves that robots can be trained and train themselves to handle an unpredictable world, and do so at high speed.
Additionally, a future table tennis competition is not completely ruled out. After all, the Sony AI team is constantly working on improving Project Ace’s gameplay. They note, for example, that the robot tends to move and strike earlier than its human opponents. Sometimes stepping back and waiting a moment can allow for a more strategic comeback.
If they solve this problem (or maybe Project Ace solves it on its own), why should the Olympics be excluded?
Follow TechRadar on Google News And add us as your favorite source to get our news, reviews and expert opinions in your feeds.
