Research On The Agent's Stratege In RoboCup

The Agent Technology is one of the most important feilds in AI. The Agent Technology includes intelligent-agent, mobile-agent, multi-agent coorperation, real-time induce-play-decision, machine learning, program design based on agent etc.The RoboCup (namely the The World Cup Robot Soccer ) is the game played by the real robot or simulation robot. The simulation league concered in this thesis is played on the standard software platform. The research field includes multi-agent cooperation, real-time, induce-play-decision, machine learning and strategy acquirement in the dynamic uncertain environment. It is the perfect platform of researching Agent Technology.In this thesis, four problems are presented. The first problem is relevant to machine learning of shoot, which we propose the shoot algorithm based on approximate space; the second one is relevant to strategy of pass, which we present the improved pass algorithm based on NLP SQP Filter method; the dynamic positioning is the third problem, which the correlative graphics is represented to solve the dynamic positioning; the last problem is concerned with designing the special language, there we discuss the basic concepts and structure of design,we get a simple implement of the generator of program frame .