Design And Implementation Of Game AI Based On Behavior Tree

In recent years,with the increasing needs of the people's spirit,Video game industry is developing fastly,new game types Have sprung up,and the game content is also constantly enriched,console game was prosperous,mobile game has opened up a huge market.For developers,the goal is always to create better content.In general,the game quality depends largely on the screen,but in today,graphics technology has been improved greatly,the people has the basic meet the requirement of the pursuit of beautiful images to more deep game content.In other words,the outside of the game is very similar to the mature graphics technology,with the modern game engine which has rapid development,developers can easily complete the gorgeous screen.Playability of the game has become a key to highlight its own characteristics,the artificial intelligence in the game is the playability of the core technology,artificial intelligence provides players a way to communicte with the game role,and it makes the experience into deeper levels of the game.This paper introduces the behavior tree which is the current mainstream game artificial intelligence technology,including its forerunner finite state machine technology,behavior tree nodes and tree technology based on machine learning.On this basis,the following results are obtained:(1)this paper studies the research significance of game artificial intelligence in game development and the research status at home and abroad;The composition and characteristics of game artificial intelligence are introduced in detail,the decision-making logic in game artificial intelligence is studied,and the types and shortcomings of the existing decision logic are analyzed.(2)based on the existing behavior tree model,this paper designs a behavior tree model based on Saras algorithm,which is used to optimize the creation of behavior tree.the simulation experiment that Use general behavior tree with Saras algorithm comparison and the behavior of the tree,shows individuals using Sarsa algorithm tree ultimately survival rate is much higher than using the general behavior of individual tree,and proves that Sarsa algorithm for the superiority of the normal behavior tree.(3)This paper designs a set of behavior tree development process based on the above algorithm,and designs and implements visual editor,XML rule definition and behavior tree library.The visual editor uses WPF technology to follow the MVVMdesign pattern so that non-technical background person can configure relevant role logic directly in the editor;Use XML to define a set of rules to describe the behavior tree;The behavior tree library provides the behavior tree basic data structure,behavior tree operation environment and behavior tree expansion code.(4)in this paper,use a mobile phone network game <xue ling jue> as the background,design and implement a set of tree based on the behavior of game AI,the game is an action role playing online games,the author is mainly responsible for client development,and complete development of several important modules.,including combat system,world system and skill system,these system are the core of the game,and also are the foundation of the AI design.The game AI base on the behavior tree technology,which guarantees its extensibility and stability.