Robotic table tennis showed up as a new event in the second World Humanoid Robot Games that kicked off Saturday in Beijing.
Featuring 12 teams from around the world, the sport event tests the abilities of the robots to see, calculate, and hit the ball.
In addition to testing autonomous control, algorithm design, and engineering capabilities, another major challenge for the teams is bridging the gap between simulations and the real world.
The robots make all decisions autonomously using AI, which makes table tennis one of the most challenging events.
They need to determine their return strategy, choose between forehands and backhands, and make other decisions in less than 0.3 seconds.
In order to train the robots, teams have them play against human opponents, and use motion-capture cameras to record the trajectory of the specially designed balls.
The data is then used to help the robots learn and improve.
Xu Kaiyang, captain of the joint team from Shanghai Jiaotong University and the Shanghai Innovation Institute, explained the challenges faced while training the robots.
"The data collected from human athletes can vary from team to team. We normally train in Shanghai, so when we came here to Beijing, we first needed to measure things like air drag coefficient. The motion-capture equipment here may also be different from what we use in Shanghai," said Xu.
"The hardest part is that the robot has to be trained in simulations first. The results may be good in simulations, but once you bring it into the real world, there can be a gap between the two. In the short term, we want the robot to get every return on the table and improve its return strategy. Right now, it's still at a beginner's level. In the long term, we want to see if we can train it to reach the level of a human athlete who has had one or two years of systematic table tennis training," he said.
This year's World Humanoid Robot Games brought together more than 2,000 robots from 600 teams to compete in 51 events that push the limits of humanoid technology.
The athletic competitions have introduced new high-intensity events including long jump, weightlifting, tug-of-war and table tennis, all of which serve as ultimate tests of a robot's "brain," "cerebellum," physical structure and core components.
In addition to athletic events, the Games also feature scenario-based competitions spanning nine fields, including home services, hospitality and logistics, demonstrating the application potential of humanoid robots in everyday production and life.
Table tennis included as new event in World Humanoid Robot Games in Beijing
