Reinforcement Learning Soccer Teams with Incomplete World Models

We use reinforcement learning (RL) to compute strategies for multiagent soccer teams. RL may pro t signi cantly from world models (WMs) estimating state transition probabilities and rewards. In high-dimensional, continuous input spaces, however, learning accurate WMs is intractable. Here we show that incomplete WMs can help to quickly nd good action selection policies. Our approach is based on a novel combination of CMACs and prioritized sweeping-like algorithms. Variants thereof outperform both Q( )-learning with CMACs and the evolutionary method Probabilistic Incremental Program Evolution (PIPE) which performed best in previous comparisons.

• Type: Journal
• Year: 1999
• Where: AROBOTS
• Authors: Marco Wiering, Rafal Salustowicz, Jürgen Schmidhuber
• Keywords: AROBOTS 1999, Continuous Input Spaces, Multiagent Soccer Teams, State Transition Probabilities