Progress in Learning 3 vs. 2 Keepaway

Gregory Kuhlmann and Peter Stone
In RoboCup-2003: Robot Soccer World Cup VII, 2004.

Abstract: Reinforcement learning has been successfully applied to several subtasks in the RoboCup simulated soccer domain. Keepaway is one such task. One notable success in the keepaway domain has been the application of SMDP Sarsa(lambda) with tile-coding function approximation. However, this success was achieved with the help of some significant task simplifications, including the delivery of complete, noise-free world-state information to the agents. Here we demonstrate that this task simplification was unnecessary and further extend the previous empirical results on this task.

@InCollection(LNAI2003-keepaway,
  author="Gregory Kuhlmann and Peter Stone",
  title="Progress in Learning 3 vs. 2 Keepaway",
  booktitle="{R}obo{C}up-2003: Robot Soccer World Cup {VII}",
  editor="Daniel Polani and Brett Browning and Andrea Bonarini and Kazuo Yoshida",
  publisher="Springer Verlag",
  address="Berlin",
  year="2004"
)