Ping-pong appears to be the preferred sport for technology companies to exhibit their robotic products. For instance, Omron, a Japanese company, garnered attention several years ago with its ping-pong automaton, which was capable of maintaining a rally with a human player while simultaneously showcasing the company’s sensor and control technology.
Additionally, the recent advancements in artificial intelligence (AI) have resulted in the emergence of more sophisticated autonomous ping-pong players that are capable of challenging even the most accomplished human players.
Consider this remarkable endeavour by the engineers at Google DeepMind. The team has developed a “solidly amateur human-level player” that combines AI and an industrial robotic arm, complete with a bat, as detailed in a recent paper titled “Achieving Human Level Competitive Robot Table Tennis.”
The AI-powered robot is depicted in a video (top) as it executes a variety of backhand and forehand strokes by making rapid decisions. It is also capable of returning shots that are served with light spin, indicating that it has the capacity to read the spin of the ball and modify the way it hits the ball accordingly.
Additionally, it is capable of accommodating balls that are struck at high and low velocities from all angles of the table, as well as those that are struck from a significant height after a significant bounce off the table. Truly, it is extremely remarkable.
The researchers stated in the paper that the robotics research community’s north star is the attainment of human-level speed and efficacy on real-world tasks. “This work advances the objective by introducing the first learnt robot agent that achieves amateur human-level performance in competitive table tennis.”
The robot defeated human beginners in all of the games it played, and intermediate players in 55% of the games it played. This was achieved through a succession of test encounters.
Nevertheless, it is evident that there is a significant amount of room for development, as it was defeated in all of its games by more experienced players. In total, the robot emerged victorious in 45% of the 29 events it participated in.