Heuristic search applied to abstract combat games

  • Authors:
  • Alexander Kovarsky;Michael Buro

  • Affiliations:
  • University of Alberta, Edmonton, Alberta, Canada;University of Alberta, Edmonton, Alberta, Canada

  • Venue:
  • AI'05 Proceedings of the 18th Canadian Society conference on Advances in Artificial Intelligence
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Creating strong AI forces in military war simulations or RTS video games poses many challenges including partially observable states, a possibly large number of agents and actions, and simultaneous concurrent move execution In this paper we consider a tactical sub–problem that needs to be addressed on the way to strong computer generated forces: abstract combat games in which a small number of inhomogeneous units battle with each other in simultaneous move rounds until all members of one group are eliminated We present and test several adversarial heuristic search algorithms that are able to compute reasonable actions in those scenarios using short time controls Tournament results indicate that a new algorithm for simultaneous move games which we call “randomized alpha–beta search” (RAB) can be used effectively in the abstract combat application we consider In this application it outperforms the other algorithms we implemented We also show that RAB's performance is correlated with the degree of simultaneous move interdependence present in the game.