Hierarchical Pathfinding and AI-Based Learning Approach in Strategy Game Design

Abstract

Strategy game and simulation application are an exciting area with many opportunities for study and research. Currently most of the existing games and simulations apply hard coded rules so the intelligence of the computer generated forces is limited. After some time, player gets used to the simulation making it less attractive and challenging. It is also costly and tedious to incorporate new rules for an existing game. The main motivation behind this research project is to improve the quality of artificial intelligence- (AI-) based on various techniques such asqualitative spatial reasoning(Forbus et al., 2002),near-optimal hierarchical pathfinding(HPA*) (Botea et al., 2004),and reinforcement learning(RL) (Sutton and Barto, 1998).