Design And Implementation Of Real-Time Decision-Making Game Agent Based On Distributed And Semi-Physical Simulation

With the continuous development and improvement of information technology such as artificial intelligence and electronic science,agent technology has been widely applied to complex system modeling and problem solving.Under the premise of various constraints,the multi-agent system can maximize the use of distributed system resources and output the strategy in real time according to the state data for different task objectives.In an increasingly complex and rapidly changing dynamic environment,to ensure a smooth decision-making process of the agent system,it is necessary to consider the decision-making information management problem that contains various computing and communication resources in the complex system.In addition,to ensure the construction of the cooperative ability of the distributed system,it is important to establish an information sharing channel to reduce the limitations of single agent observation.Based on the research at home and abroad,thesis analyzes the functional requirements of real-time decision-making agents,and establishes an agent system with decision information processing and real-time decision-making capabilities.The main work of thesis includes the following aspects:Firstly,according to the different composition of the decision-making information in various simulation environments,thesis defines the basic data types,and models the hierarchical decision-making information resources.Otherwise,it also designs and implements a variety of decision-making information processing services.Through the simple and efficient information operation interface,the redundancy of information transmission is reduced,and the unified management ability of decision-making information is provided for the agent system.Secondly,aiming at the problem of multi-agent cooperation in distributed systems,this thesis focuses on the key part of information sharing between agents,and constructs an information transmission method between agents based on routing information table,which provides a basic message path for information transmission between agents,and designs and implements the decision information distribution agent.Thirdly,to meet the real-time decision-making needs of the system,this thesis studies the utility function construction method of the agent.By constructing the utility function based on the light-source tree to meet the calculation requirements of individual decision-making,the utility information sharing is used to meet the needs of collaborative decision-making,and the design and implementation of agents with different roles are carried out in the game scene.Finally,to test the function of each software module and the decision-making effect of the agent,this thesis builds a distributed semi-physical simulation system,and the experiment was conducted in the Pac-Man game scene.The results show that the agent system has the ability to quickly process decision information and can efficiently complete scheduled missions.