The AI robot from Google DeepMind can evaluate scenarios while playing table tennis, even if it’s not a professional match.
Google DeepMind described in a paper that was posted on Arxiv last week how the robot uses some video to challenge players at various skill levels.
DeepMind claims that although the robot is shaped like an arm, which is typical in manufacturing facilities, it has an AI system and camera to assess circumstances. While it isn’t yet capable of serving the ball directly, the robot is trained to perform both simple tasks like returning the ball and more difficult ones like organizing play and strategizing.
The AI robot “combated” a variety of opponents with different playing styles in addition to the input data, gathering information to improve and modify its tactical strategy. The robot won 13 out of 29 matches using the 3D printed racket; it won all matches against “beginners,” 55% of matches versus “intermediate” players, and lost all matches against “high-end” opponents.
In an interview with TNW, a DeepMind spokesperson stated that the project’s outcomes indicate a significant advancement toward the objective of building robots with “human-level” performance and speed for everyday jobs. They will first assist the entertainment industry before being used in more intricate domains.
Table tennis is not the only sport used by Google DeepMind to train its AI system. The Japanese business Omron has previously carried out a similar action. Table tennis is seen to be an appropriate activity for training and assessing the abilities of intelligent robots because of its requirements for hand-eye coordination, strategic thinking, speed, and flexibility. synthetic intelligence.
