Modeling And Application Of Game AI Based On Fuzzy-logic Inference Behavior Tree

Behavior tree is a commonly used technique for game designers to describe in-game agent behaviors.With the complexity of decision logic,readability and decision-making accuracy of behavior tree will be reduced rapidly,which has a major impact on the design work of game designers.Therefore,this paper studies the optimization for readability and decision-making accuracy of behavior tree.The research work in this paper is mainly in the improvement of behavior tree,which is embodied in two aspects:(1)Improved behavior tree based on premise node: the improved behavior tree increases the information and functionality of node and simplifies the organization of behavior tree by using premise node instead of condition node to determine the conditions.Theoretical analysis and case studies show that the improved behavior tree effectively improved readability and efficiency.(2)Improved behavior tree based on fuzzy logic combined with premise node: the improved behavior tree can calculate and analyze execution expectation of node by using the fuzzy logic to describe the fuzzy prerequisite,which improving the decision-making accuracy of behavior tree.The results of experimental analysis and game test show that the improved behavior tree has a better performance in improving the intelligence level of in-game agents.Finally,this paper proposes a fuzzy-logic inference behavior tree,which is applied to the ARPG game AI system,which verified its validity in design of decision logic.For experienced game designers,the improved behavior tree can provide them with more technical options.For general game designers,the improved behavior tree which requires less experience and ability,is a better technical option.