next up previous
Next: The Simulator Up: Towards Collaborative and Previous: Towards Collaborative and

Introduction

Soccer is a rich domain for the study of multi-agent learning issues. Teams of players must work together in order to put the ball in the opposing goal while at the same time defending their own. Adaptive learning is essential in this task since the dynamics of the system can change radically as the opponents' behaviors change. The players must be able to adapt to new situations.

Not only must the players learn to adapt to the behavior of different opponents, but they must learn to work together. Soon after beginning to play, young soccer players learn that they cannot do everything on their own: they must work as a team in order to win.

We are using a robotic soccer system to study both adversarial and collaborative multi-agent learning issues. We have had some initial success at using neural networks (NNs) for learning a low-level behavior for a single-agent task in a simulator [Stone & Veloso1995]. We are currently working on improving this behavior while simultaneously extending the task to capture these two types of multi-agent issues.

Here we briefly describe the experimental framework along with the initial learned behavior. We then discuss some of the issues that are arising as we extend our task to require collaborative and adversarial learning.



Peter Stone
Wed Jan 17 10:57:03 EST 1996