Google's "Deep Thinking" company recently announced that its research and development team has developed a table tennis robot that can reach the level of human amateur table tennis players in competitions. The R&D team posted on the preprint website arXiv that this is the first learning robot intelligent agent to reach the level of a human amateur player in table tennis competitions. Its main body is a six axis robotic arm that can move forward, backward, left, and right through a bottom slide rail. In 29 matches against humans, robots won 13 with a win rate of 45%. The opponents are all human players that robots have never seen before, with skill levels ranging from novice players to advanced players. Researchers say that robots lost all matches against advanced players, but won all matches against junior players and 55% of matches against intermediate players. To achieve human level speed and performance, the R&D team has adopted a layered and modular strategy architecture, allowing robots to not only master "low-level skills" such as forehand topspin and backhand push, but also develop strategies through a "high-level controller" equivalent to the brain. During the competition, the 'advanced controller' can develop the best skill plan based on the actual situation of the competition, the robot's own abilities, and the opponent's abilities. After the game, the robot can also analyze battle data and continuously improve its skills. Researchers say that this robot still has many shortcomings, such as weak backhand play, and is not good at dealing with fast balls, high, low, or strongly spinning balls. They will attempt to further enhance the performance of the robot by improving control algorithms, optimizing hardware, and other methods. (New Society)