AI Research And Application In Online Game

AI system usually is a significant part of game software. The purpose of using AI techniques is to utilize AI research achievements and emulation techniques for creating virtual characters and objects with intelligent behavior. And it has become the fundamental techniques in improving game quality and user experience. With the advance of online game industry, the game development has become works under platformization, the emphasis on game personality and user interactive experience makes the research and application of AI related technique a very important topic.As a part of game software, an AI system has to coordinate with other modules of game engine. It can be divided into several major components: Path-Finding and Movement, Script System, Sensing and Drive Layer, Decision System. The Decision System is the core of AI System and other parts are support components. For implementing an efficient and stabilized AI system and so as to better serve game designing, the internal function of each sub-systems and the cooperative relationship between them need to be researched, and strategies for developing those modules need to be concluded.This thesis researches AI related techniques of video game console platform, considering AI decision characteristics under network environment and design requirement of Monster Hunter game, it presents a game AI system with behavior tree decision system, multi-layer path-finding structure based on NaviMesh, and static environment sensing combined with dynamic sensing. The Decision system is based on fundamental behavior tree decision way but its framework and workflow are extended and redesigned for enhancing the reusability of AI logic and interactivity between the AI decision process and other systems. The major work includes: Introducing reference node into behavior tree for realizing AI logic modularization during modeling and editing phase, then assembling the logic components into final complete run-time behavior tree structure, thus the AI logic modularization method improves the development efficiency of AI system; Designing behavior tree running in two ways, one is running across frames which based on AI object, another is running confined to single frame based on event reaction, with this method different AI logic functions are separated and organized into behavior trees; Introducing AI object based blackboard mechanism for sharing data across different behavior trees and sharing info between behavior tree and external system. The Path-Finding and Movement system was implemented in multi-layer path-finding structure which considers the partition battle characteristics of Monster Hunter game and running efficiency over network environment. The environment sensing process was implemented combining two ways, one is static environment sensing which binding static info point with NaviMesh node, another is dynamical environment sensing which relying on grid data of battle partition.The above AI system scheme has been applied in Monster Hunter Online game development, this system meet the requirement of game designers for designing various AI genres, and the data driven based behavior tree decision method also improves the efficiency of AI iteration development. In addition, the multi-layer path-finding structure, the division of static sensing and dynamical sensing ensures the AI system running in high efficiency under network environment.