In sign #123,265 that the robot apocalypse is almost upon us, an autonomous AI-powered robot can now defeat top-level human competitors in table tennis.
On the plus side, at least for us humans, the robot, named Ace, did struggle a little bit against the pros. On the negative side, it prevailed in most of its matches against experienced non-professional table tennis players. In fact, serving against several top players, Ace scored 16 aces.
This is the first time a robot has achieved “expert-level play in a commonly played competitive sport in the physical world,” Sony AI said in a statement.
As the authors of a study analyzing Ace’s table tennis skills wrote in the journal Nature, “Artificial intelligence (AI) systems now challenge or surpass human experts in many computer games. Physical and real-time sports such as table tennis, however, remain a major open challenge because of their requirements for fast, precise and adversarial interactions near obstacles and at the edge of human reaction time. Here we present Ace, to our knowledge, the first real-world autonomous system competitive with elite human table tennis players.”
Yipee?
How can a robot beat the world’s best human table tennis players?
Ace uses three key developments in autonomous robotics to achieve this feat. First, it uses “event-based sensors,” which means that to track the table tennis ball’s path, the robot focuses on specific areas in the images its cameras capture, such as those that indicate changes in motion or brightness.
The robot then develops its table tennis abilities through “model-free reinforcement learning.” The robot received thousands of hours of training during this process.
Lastly, high-speed robot hardware enables Ace to play with “human-like agility.” According to Peter Durr, Director of Sony AI in Zurich, human athletes need about 230 milliseconds to react, whereas Ace’s total latency is only about 20 milliseconds, making it in some ways even more agile than a human.
But, why?
“This research has shown that an autonomous robot can, in fact, win at a competitive sport,
matching or exceeding the reaction time and decision making of humans in a physical space,” Durr explained.
“Table tennis is a game of enormous complexity that requires split-second decisions as well as speed and power. This research breakthrough highlights the potential of physical AI agents to perform real-time interactive tasks, and represents a significant step toward creating robots with broader
applications in fast, precise, and real-time human interactions.”