Reinforcement learning soccer teams with incomplete world models

Marco Wiering, Rafał Sałustowicz, Jürgen Schmidhuber

Research output: Contribution to journalArticlepeer-review

24 Scopus citations


We use reinforcement learning (RL) to compute strategies for multiagent soccer teams. RL may profit significantly 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 find good action selection policies. Our approach is based on a novel combination of CMACs and prioritized sweeping-like algorithms. Variants thereof outperform both Q(λ)-learning with CMACs and the evolutionary method Probabilistic Incremental Program Evolution (PIPE) which performed best in previous comparisons.
Original languageEnglish (US)
Pages (from-to)77-88
Number of pages12
JournalAutonomous Robots
Issue number1
StatePublished - Jan 1 1999
Externally publishedYes

ASJC Scopus subject areas

  • Artificial Intelligence


Dive into the research topics of 'Reinforcement learning soccer teams with incomplete world models'. Together they form a unique fingerprint.

Cite this