AI BriefingRobot Table Tennis Breakthrough
·2026.04.23 06:12
Key point
Sony AI's robot Ace defeated elite players in an official-rules match.
1 / 2
Details
Sony AI's table tennis robot Ace won 3 out of 5 matches against elite players in an official-rules match.
- It lost 2 matches against 2 professional players, but overall it proved it can compete at a level competitive against humans.
- The key was spin recognition and ultra-fast reaction, responding instantly even when the ball's trajectory changed after catching the net.
- It tracks the ball's position and rotation with multiple cameras, and estimates the axis of rotation by zooming in on the logo on the ball's surface.
- Shot selection and spin response were refined through 3,000 hours of training performed in simulation.
- Ace was initially weak against slow, low-spin balls, but showed impressive performance on tricky shots.
- The research results were published in Nature, and the development team said performance has improved further since submission.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Xiaomi Open-Sources Embodied AI Foundation Model Xiaomi-Robotics-1 (4 min read)
Transformer Transformer: A Unified Model for Motion-Conditioned Robot Co-Design