Sony AI has unveiled a major breakthrough in robotics with a system that can go head-to-head with top human athletes in one of the fastest sports on the planet: table tennis.
The project, known as Project Ace, is the first real-world autonomous robot shown to compete at an elite level in a widely played sport. The research behind it was published in Nature, marking a milestone that has long been out of reach for robotics.
For years, artificial intelligence has dominated digital arenas like chess and Go. But translating that success into the physical world has been far more difficult. Table tennis, in particular, pushes systems to their limits. The ball moves at high speeds, spins unpredictably, and forces split-second decisions that leave almost no room for error.
That’s exactly why Sony chose it.
Project Ace combines high-speed vision systems, reinforcement learning, and precision robotics to track, predict, and respond to the ball in real time. Using a network of advanced cameras and sensors, the system can calculate the ball’s position, velocity, and spin almost instantly, then execute a return with human-level timing and accuracy.
In testing, the results were striking. Ace competed against both elite and professional players under official rules and managed to win multiple matches against elite opponents. It also demonstrated a strong ability to handle spin, one of the hardest elements of the game, while reacting to unpredictable shots like net deflections.
The performances didn’t stop there. In follow-up matches conducted after the initial research, the robot showed continued improvement, notching wins against professional players and playing with faster, more aggressive rallies. The system’s ability to adapt and improve highlights a key shift in robotics, where machines are no longer just repeating programmed actions but learning dynamically from experience.
The implications go well beyond sport. Table tennis serves as a proving ground for real-time interaction in the physical world, where perception, decision-making, and movement all have to happen in milliseconds. Solving that problem opens the door to applications in areas like manufacturing, logistics, and other environments where speed and precision are critical.
Project Ace builds on earlier work by Sony AI, including its success in virtual environments with racing AI systems. But this time, the challenge was real, physical, and unpredictable.
Join the Discussion