The Research Of RoboCupSoccer Simulation 2D

The Multi-Agent System is one of important research directions in artificial intelligence in recent years. As an example of the Multi-Agent System, RoboCup will provide a standard to encourage the researchers of artificial intelligence and the intelligent robot domain. So they can fully use each kind of effective technology to obtain the better solutions. RoboCup simultaneously provides a software platform to conduct the software simulation research.Based on the fully understanding of Robocop simulation competition model— Soccer Server's operational mechanism, this article analyzes the simulation 2D contest rule and the competition process. Based on the TsinghuAeolus2002's first floor code, this article establishes a simulation robot football team ZzjyxyFS. First, it introduced the ZzjyxyFS's whole design of structure; The ZzjyxyFS agent uses three structural, including first floor, technical level and decision-making strata. And the first floor is responsible to communicate with SoccerServer and to obtain the vision and the sense of hearing information from SoccerServer, and it can renews and maintains a relatively precise world model which other levels need; The technical level mainly be used to define the player agent's individual technology and all of individual movement (many of which are compound movement), like intercepts a pass, kicks a ball, obstructs a individual, stares at the human and so on ; The decision-making level is responsible to carry on the real-time decision-making according to the competition situation , which is the core part of ZzjyxyFS agent structure. Next, this paper discusses four types of the Multi-agent cooperates, cooperated we in view of the RoboCup Multi-agent level to analyze the neural network technology in the ZzjyxyFS agent system design application, including designs and trains the success ratio each kind of basic model and the description intercepted a pass; this paper uses the fuzzy reasoning mechanism in passing a ball in the line appraisal in the team and the dribbling decision-making , and unified shoots the movement to form player's ball control strategy. Finally, it carries on the simulation and the experiment based on the simulation competition system to ZzjyxyFS, and simulates the competition process. From the simulation's result, its policy-making strategy is feasible. ZzjyxyFS has obtained the good result in 2006 Chinese Robot soccer simulation 2D competition.