Soccer Robots Local Passing And Tactics Cooperation

The theoretical research and practical application of humanoid robot has become a hot spot in the research of artificial intelligence.With the continuous improvement of computing power of computer or computer cluster,machine learning has become a hot spot and focus in the research and application of the academia and the industry.First of all,the basic problems of robot research include the action design and optimization of the individual robot.The soccer robot can only carry out the study of the upper level strategy only after mastering the flexible and fluent behavior.Smooth walking plays a vital role in the robot soccer games.This paper presents an algorithm for the gait design of a biped robot based on a three-dimensional linear inverted pendulum model.The simplified three-dimensional linear inverted pendulum model is applied to the robot's leg,and then through the kinematics equation derived function centroid and time with the parameters of the robot gait design reduced to each walk cycle,the method is extended to design with the steering gait and the hierarchical gait reinforcement learning method is used to optimize the designed gait parameters.The flexibility and mobility of the walking and turning of the robot have been improved in an all-round way.Secondly,the study of kicking action is an essential one for the robot competition which is the carrier of football,in order to solve this problem,a kicking method based on Bessel curve is proposed.The method is mainly composed of the trajectory planning module and the inverse kinematics module: The trajectory planning module is responsible for the application of the Bessel curve interpolation between the foot and the kickball point to generate the continuous trajectory;The inverse kinematics module updates the angle values of each leg's legs in real time based on the generated trajectories,so as to ensure that the robot can perform the complete kicking action and the 23 behavioral parameters necessary for practical application are optimized.The method improves the accuracy and the strength of the player to kick the ball.Finally,the collaboration and antagonism of multi agents belongs to the upper level strategy.For achieving a good teamwork,this topic is based on the skillful individual motion technology of a single robot,and studies the strategy of passing the ball.The KeepAway training model and the pass area determination module constitute the local collaboration tactics.KeepAway is a typical intensive training model.In the limited range,a well designed pass plan is trained to get the stable pass strategy and the longest holding time.The pass area determination module starts from the passing objects,studies the identification of passing objects and the information transmission between the passing objects,so as to determine the effective passing area and plan the best passing route.