Learning to Dribble on a Real Robot by Success and Failure

This video describes how a neural-network-based RL controller learns the challenging task of ball dribbling directly on a Middle-Size robot. The learned behaviour was actively used throughout the 2007 RoboCup world championship tournament in Atlanta, where the robots won the first place. This is part of the Brainstormers project, which aims to develop an intelligent control architecture for a soccer playing robot that is able to learn increasingly complex behaviours from scratch.