Study On The Defensive Action And Movement Of Simulation 2D Soccer System

Robocup 2D simulation platform is a dynamic multi-agent countermeasure system.In the simulation platform,the action selection of player agent in each cycle will directly determine the team's offensive and defensive capabilities,how to cooperate with each other,how to be more accurate,fast to reach their target point of attack or defense of the players in the process are prerequisites for all effective strategies.This paper takes the selection of agent action in defensive task and the player movement in the conversion as the focus on the basis of the design of triangulation.The research substances are as follows:First,the Monte Carlo tree search algorithm is introduced into the 2D simulation,it uses player agent to define nodal point of game tree,and players' choices of movement are regarded as transition among nodes.For defensive works,it builds model of Monte Carlo.It utilizes polar coordinates system to make region segmentation,also makes combination of Q learning and Upper Confidence Bound Apply to Tree of Monte Carlo exploring method to train the team players.While using evaluated number of movement of training results as optimized competition codes in order to get a better action selection strategy.Secondly,this thesis proposes a scalable role allocation method based on time minimization in the problem of coordinated motion of distributed agents,deeply analyzes and compares the different implementations of the method,and applies it to the formation of team offensive and defensive conversion and local coordination position problem in the process of attacking and defending for players in the 2D platform.Model the position problem of player group,so the player runs is more scientific and sensitive,and reduce unnecessary mistakes.Finally,By defining the state of the attack and defense as the root node in the Monte Carlo tree,combining the time minimization role assignment method with the intelligent group defense experiment,then through the game data to prove the scientific nature of the method.