Sony's latest innovation, an autonomous robot named Ace, has showcased its capabilities by winning three matches against elite table tennis players. This development signals a significant milestone in the field of physical AI, demonstrating that such systems can now compete at expert levels in complex sports.
Equipped with advanced sensory technology, Ace utilizes nine pixel-sensor cameras to track the ball's position in three-dimensional space, alongside systems that assess its velocity and spin. The robot's decision-making is powered by a model-free reinforcement learning control system, allowing it to adapt in real-time without preprogrammed instructions.
Peter Dürr, director of Sony AI, highlighted the complexity of table tennis, noting its demands for quick decisions and agility. In a series of tests, Ace scored 16 points while serving, compared to 8 points by its elite competitors. This performance not only demonstrates Ace's potential but also hints at future applications of physical AI in various fields.
Peter Stone, chief scientist at Sony AI, emphasized that achieving expert-level operation opens the door to numerous real-world applications previously deemed unattainable.