When it comes to demonstrating the prowess of robotics in the field of sports, ping-pong seems to be the top choice for tech companies. Take the Japanese firm Omron, for instance. Several years ago, it made waves with its ping-pong robot that could effortlessly engage in a rally with a human player, simultaneously flaunting the company’s cutting-edge sensor and control technology.
And with the recent meteoric progress of artificial intelligence (AI), we are now witnessing the emergence of more sophisticated robotic ping-pong players that might soon pose a serious challenge to even the most skilled human players.
Consider this remarkable achievement by engineers at Google DeepMind. In a new paper titled “Achieving Human Level Competitive Robot Table Tennis,” the team announced that they have crafted a “solidly amateur human-level player” by integrating AI with an industrial robotic arm equipped with a bat.
A video (at the top) showcases the AI-driven robot making rapid decisions to execute a variety of backhand and forehand strokes. Notably, it can also return shots served with a light spin, demonstrating its ability to discern the spin on the ball and adjust its hitting method accordingly. It is capable of handling balls coming at it at both high and low speeds, from all areas of the table, as well as balls arriving at a considerable height after taking a significant bounce on the table. Truly, it is highly impressive.
“Achieving human-level speed and performance in real-world tasks is a guiding star for the robotics research community,” the researchers stated in the paper. “This work takes a significant step towards that objective and presents the first learned robot agent that attains amateur human-level performance in competitive table tennis.”
Through a series of test matches, the robot emerged victorious in 100% of the games it played against novice human players and 55% against intermediate players. Nevertheless, there is evidently ample scope for improvement as it lost all its games against advanced players. Overall, the robot won 45% of the 29 games it contested.