Gait Planning In RoboCup Soccer Simulation Competition

The research of robot has come to humanoid stage from just simulating human walking. The RoboCup (The Robot World Cup) organized the game of humanoid soccer robot. Robot soccer related technology has a broad foreground and prepared for robotics application in industry and commerce. Since humanoid robot simulates the locomotion method from human being, it suits to help people human-friendly or cooperate with people in the daily life and works. This paper studied the gait planning of humanoid robot under robot soccer simulation platform.In the D-H coordinate, direct kinematic model is built, and the solution of robot vision mode is given, so direct kinematic could be solved on line. In accordance with inverse kinematic problems, a new inverse kinematic algorithm, whose input are pose of foot and horizontal coordinates of COG, is introduced for humanoid gait planning. The problem is converted to multivariate function optimization and conjugate gradient descent method is used to find the solution of multivariate function. In the robot program, matrix operation is optimized to improve calculation speed.The usual stability criterion of humanoid is outlined and two gait planning method based on foot steps and COG is put forward. The first method, driven by the target of two foots, generate trajectories of footstep and COG. The results of experiment show that this method could be used when is velocity of robot is low, but does not fit the simulation competition. The second method integrates offline key frame and online adjustment to achieve real time control. To solve the transition of key frame, the shortest path algorithm of directed graph is used. To realize the top decision-making of gait planning, Finite state machine is used. The results of experiment show that the calculation speed of method is fast and could realize real time gait adjustment under simulation platform with noise.