Font Size: a A A

Design And Implementation Of Football Robot Simulation System Based On Multi-agent

Posted on:2009-11-13Degree:MasterType:Thesis
Country:ChinaCandidate:H M SunFull Text:PDF
GTID:2178360272490919Subject:Software engineering
Abstract/Summary:PDF Full Text Request
With the development of computer technology,research of the MAS(Multi-agent System) has become an important content of artificial intelligence research and computer field.RoboCup(Robot World Cup),namely Robot World Tournament,Robot World Cup in recent years has received broad attention in one area of AI study and have important theoretical significance and broad application prospects. RoboCup is A typical platform of MAS research.It passes the problem providing a standard to researcher.Encourage Artificial Intelligence researcher research about a problem under a real time synchronization,antagonism environment having noise,decision-making and cooperation of many intelligence bodies.This paper is established in RoboCup emulational 2D-group,design the stations Strategy and ball strategy,and validate our strategy by simulate match.The Station Agent-processing strategy which based on bottom of the world on the basis of the model is made by analysis of the current state of competition. On this paper,we do a Design and Research of the 2D Simulation Group's client's intelligent-by learning Artificial intelligence,neural networks,Soccer Server(2D) Simulation environment and the experience of football experts.In the design,we use BP neural network to achieve the basic definition of the point of camps. Transform the state between defensive and offensive,then complete Research and Implementation of the stations offensive strategy and defensive strategy stations.In RoboCup simulation game,each player is an agent.Team strength and weakness largely reflects the skills of the player agent moves designed superiority and inferiority.Senior tactical strategy will ultimately be transformed into a series of movements.Based on the algorithm analysis and improvement of neural networks and fuzzy systems,We achieve the ball strategy, intercept strategy,and goalie strategy inside the RoboCup.We debug the basic skills online,and it mostly meet our expectations,has a good real-time.
Keywords/Search Tags:RoboCup, Multi-agent, Stations Strategy, Action Strategy
PDF Full Text Request
Related items