Sony AI’s table tennis robot Ace humbles players at national level
Sony AI's table tennis robot Ace defeated three of five national-level players, but lost to two professionals. The robot's AI system was trained using deep reinforcement learning and can analyze the ball's position and rotation with 12 cameras.
A research team led by Sony AI developed an artificial intelligence-based robot named Ace, capable of playing table tennis at a national level. Ace defeated three of five players who have competed in national tournaments. The robot's 'vision' system, equipped with 12 cameras, analyzed the ball's position and rotation. The racket arm's motion was trained using deep reinforcement learning, enabling Ace to employ 15 different serves and sustain rallies. Matches were refereed by a human and played under standard rules. Ace's bouts against human opponents tested its capabilities, with rallies averaging five strokes. The team's findings were published in the British scientific journal Nature on April 23.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.