Fuzzy Theory Based Single Belief State Generation for Partially Observable Real-Time Strategy Games

As the basic problem of the real-time strategy (RTS) games, AI planning has attracted wide attention of researchers, but it still remains as a huge challenge due to its large searching space and real-time nature. The situation may get worse when the planning in RTS games is implemented under a partially observable environment considering the existence of the fog-of-war. Given the recorded past positions of an agent, it would be helpful if the targets' next position can be predicted based on the recorded data since this will increase the certainty of the target. Therefore, this paper proposes a fuzzy theory-based single belief state generation method named FTH to do what based on multi-layer information sets extracted from the history position information. Besides, we incorporate the FTH generation method into adversarial hierarchical task network repairing (AHTNR) planning algorithm, which can be used for the prediction of the unit's position and task planning. Finally, we carry out an empirical study based on the mu RTS game and validate its effectiveness by comparing its performance with that of other state-of-the-art algorithms.