Sign In

Communications of the ACM

ACM TechNews

Brain-Training For Baseball Robot

Artist's conception of a baseball-playing robot.

Researchers have implemented a model of the part of the human brain responsible for fine motor control on a graphics processing unit connected to a small humanoid robot, to determine whether it could teach the robot accurate timing in hitting a pitched ba


Researchers at the University of Electro-Communications (UEC) in Japan have developed a model of the part of the brain that is responsible for fine motor control.

The model of the cerebellum, which consists of more than 100,000 neurons, is a large-scale version of a so-called spiking network model. The team implemented the artificial cerebellum on a graphics processing unit, and then connected the model with a small humanoid robot to test whether or not it could teach the robot accurate timing.

UEC's Tadashi Yamazaki and the Okinawa Institute of Science and Technology's Jun Igarashi sought to train the robot to hit a ball bowled in real time by a bowling machine. Through repeated practice over time, the robot learned when to raise a bat in order to hit a flying ball.

The researchers believe the real-time cerebellum could be used to teach and train robots for different tasks in the future.

From University of Electro-Communications (Japan)
View Full Article


Abstracts Copyright © 2015 Information Inc., Bethesda, Maryland, USA


No entries found