Sony AI's Ace robot beats pro table tennis player, runs on ROS 2
First robot to defeat a pro using 10.2ms perception and 1kHz control loop.
Sony AI's Ace robot has achieved a historic milestone by defeating a professional table tennis player, earning the cover of Nature. This is the first time a robot has outperformed a human in such a fast-paced, dexterous sport. Ace operates with a stunning 10.2 ms perception latency using event-based cameras to track a ball moving at 20 m/s with spin exceeding 9,000 rpm. Its control loop runs at 1 kHz, enabling split-second decision-making and swing execution, with the ability to return the ball at speeds up to 19.6 m/s. The entire system is powered by ROS 2, proving that the robotics middleware can handle the most demanding real-time tasks in physical AI.
For professionals, Ace demonstrates that ROS 2 is production-ready for high-speed, closed-loop control systems—previously the domain of custom real-time stacks. The implications extend beyond table tennis: warehouse sorting, autonomous drone racing, and surgical robots can all benefit from this proven architecture. Sony AI's success validates ROS 2 as a scalable backbone for next-generation robotics, lowering the barrier for teams to build similarly responsive systems without reinventing the wheel.
- Ace is the first robot to beat a professional table tennis player, featured on the cover of Nature.
- Perceives a ball at 20 m/s with 9,000+ rpm spin in 10.2 ms using event-based vision and a 1 kHz control loop.
- Runs entirely on ROS 2, proving its real-time capability for extreme physical AI tasks.
Why It Matters
Validates ROS 2 for real-time, high-speed robotics, opening doors for dexterous automation in warehouses, surgery, and more.