A Game AI Scheme Basedon Utility Theory Is Oriented To Multi-game Characters

With the continuous improvement of the quality of video games,the importance of the playability of video games has become increasingly prominent.However,the basis of all playability depends on a perfect AI performance.Therefore,major video game companies are gradually increasing the research and development of AI.In recent years,with the popularity of competitive game market,most electronic game companies have competitive game research and development projects.However,competitive game is a standard content-based expansion game,which needs to extend its life cycle by constantly creating new characters.As a result,the development cycle of AI is getting longer and longer.Many games have been developed,but the AI system still cannot meet the demand,which is a serious waste of development funds.In the process of game development,typically we use a Finite State Machine or Behavior Tree scheme to build the AI,but the Finite State Machine is difficult to expand,behavior tree upfront development costs are high,and both of them have weak batch processing ability and poor readability when dealing with multi-game characters AI development.These issues are holding back development efficiency.This paper aims to study a new game AI design method to avoid the defects of traditional AI schemes,so as to improve the development efficiency,and implement the design with JAVA language.Finally,the design method of this topic is applied in a competitive mobile game of SohuChangyou.The results show that the scheme,on the premise of meet the performance requirements,is able to make a game characters correctly express the intention of the designers,and greatly reduce the workload of programmers.The work of AI logic filling is returned to the designer,and the difficulty of AI logic filling in complex environment is reduced,which greatly improves the efficiency of game development.