Mention AI and robots to most people, and they’ll typically think of end-of-world scenarios involving lasers and glowing red eyes. Not so for Google DeepMind, who have made a significant step forward in robotic AI technology by training a system to beat the average person at table tennis rather than focusing on the global eradication of humanity.
Training the System: The Initial States
To start things off, Google DeepMind built a database of all the initial states a table tennis ball could have, such as position, speed, and spin. From there, the robot arm practiced various movements, getting used to switching between forehand and backhand grips, applying topspin, and so on.
Meet our AI-powered robot that’s ready to play table tennis. 🤖🏓
It’s the first agent to achieve amateur human level performance in this sport. Here’s how it works. 🧵 pic.twitter.com/AxwbRQwYiB
— Google DeepMind (@GoogleDeepMind) August 8, 2024
Real-Time Adaptation Against Real Players
Then, the AI system was pitched against real players. It was designed to monitor how different people would behave and play, using that information to refine the overall algorithm. Its success rate was tracked, with the selected strategy self-adjusting in real-time accordingly.
Performance and Future Applications
Google DeepMind says that the robot played against 29 opponents, ranked into four different skill levels. After completing all the matches, it came roughly in the middle of them—essentially the same as an “intermediate amateur.” Of course, it was defeated by the better players, as aspects such as how paddle rubber affects spin are difficult to model properly in simulations.
Even so, it’s still an impressive achievement. This technology has multiple potential applications, particularly in mass-manufacturing production lines that use robots for tasks like painting or welding. These robots often struggle to cope with minor mispositions or changes in lighting. An AI system trained to react properly to these variables could help prevent production lines from stalling.