Sign In

Communications of the ACM

ACM TechNews

Robot Plays Table Tennis After 90 Minutes of Training

View as: Print Mobile App Share:
The robot returns a shot.

Jonathan Aitken at the University of Sheffield in the U.K. finds it interesting that the robot sometimes struggled to make shots because of its mechanical limitations, rather than due to shortcomings with the algorithm.

Credit: Yapeng Gao, Jonas Tebbe, and Andreas Zell

Researchers at Germany's University of Tübingen developed a computer simulation that taught a virtual robot arm outfitted with a table tennis racket how to return virtual ping pong balls.

A machine learning algorithm observed the simulation for 90 minutes, learning how the ping pong ball's path was affected by velocity and the racket's orientation.

The researchers then used the algorithm to control a real robot arm at a real table tennis table, and found the robot arm could return balls within an average of 24.9 centimeters of the target location. However, it struggled with fast, slow, and backspin shots.

Said University of Tübingen's Andreas Zell, "It's not worse than a regular human player. It's already on par with me."

From New Scientist
View Full Article


Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account