Sony's ace robot defeats table tennis players using reinforcement learning
Sony's new robotic arm, Ace, is making waves by actually beating professional table tennis players.
Ace used a court setup with nine cameras positioned around it to measure the ball's spin, Ace learns through reinforcement learning and improved during later experiments, showing just how far robotics have come.
Ace won 3 of 4 matches
Sony set up a full Olympic-sized court at its Tokyo headquarters to put Ace to the test.
With eight flexible joints for fast, precise moves, Ace faced off against four top players in December and won three out of four matches.
Michael Spranger from Sony AI shared that its focus was on smart tactics rather than just speed, which really paid off.
Michael Spranger says strategy over speed
Ace's win isn't just about table tennis: it hints at big possibilities for AI in other areas like manufacturing or any fast-changing environment.
Spranger pointed out that using strategy over raw speed marks a new direction for smarter robots in real-world tasks.