LogInThe robot won almost half of the games with humans.
Recently, Google announced a notable development in the sphere of artificial intelligence – the first AI-powered robot that can play table tennis and is claimed to acquire amateur human-level performance in this kind of sport.
As noted by Google DeepMind, the robotic system created for playing this sports game has to be skilled at relatively simple actions like returning the ball, as well as master high-level skills, such as strategizing and long-term planning. The developers gathered a dataset of initial table tennis ball states to train the robot, including information about position, speed, and spin. The training involved competing with 29 human opponents with 4 different skill levels, from beginner to advanced. The robot won almost half of the games, mainly against players with intermediate skill levels.
According to the developers, the remaining challenge concerning this work is mastering the robot's accuracy, speed, and adaptability. However, Google states that while this work focuses on table tennis, it shows how robots can learn to perform complex tasks in the real world that require physical skill, perception, as well as strategic decision-making.
Read the details here and join our 80 Level Talent platform and our Telegram channel, follow us on Instagram, Twitter, LinkedIn, TikTok, and Reddit, where we share breakdowns, the latest news, awesome artworks, and more.