论文信息 - Learning to dribble on a real robot by success and failure

Learning to dribble on a real robot by success and failure

Learning directly on real world systems such as autonomous robots is a challenging task, especially if the training signal is given only in terms of success or failure (reinforcement learning). However, if successful, the controller has the advantage of being tailored exactly to the system it eventually has to control. Here we describe, how a neural network based RL controller learns the challenging task of ball dribbling directly on our middle-size robot. The learned behaviour was actively used throughout the RoboCup world championship tournament 2007 in Atlanta, where we won the first place. This constitutes another important step within our Brainstormers project. The goal of this project is to develop an intelligent control architecture for a soccer playing robot, that is able to learn more and more complex behaviours from scratch.

[1] Martin A. Riedmiller,et al. Using Machine Learning Techniques in Complex Multi-Agent Domains , 2003 .

[2] Martin A. Riedmiller. Neural Fitted Q Iteration - First Experiences with a Data Efficient Neural Reinforcement Learning Method , 2005, ECML.

[3] Martin A. Riedmiller,et al. Effective Methods for Reinforcement Learning in Large Multi-Agent Domains (Leistungsfähige Verfahren für das Reinforcement Lernen in komplexen Multi-Agenten-Umgebungen) , 2005, it Inf. Technol..

[4] Martin Lauer,et al. Making a Robot Learn to Play Soccer Using Reward and Punishment , 2007, KI.

[5] Martin A. Riedmiller,et al. On Experiences in a Complex and Competitive Gaming Domain: Reinforcement Learning Meets RoboCup , 2007, 2007 IEEE Symposium on Computational Intelligence and Games.

[6] Martin A. Riedmiller,et al. Neural Reinforcement Learning Controllers for a Real Robot Application , 2007, Proceedings 2007 IEEE International Conference on Robotics and Automation.