The framework design of humanoid robots in the robocup 3D soccer simulation competition

Abstract

This paper presents the key framework of the simulated humanoid soccer robots designed by the Apollo3D team, which is one of the most competitive teams in the RoboCup 3D competition. First, we briefly introduce the architecture of the agent client. Then, we focus on the three major components of the framework, which include a particle filter and Kalman filter based localization system, a dynamic footstep planner using sequence approximation, and a hierarchical decision making mechanism. We compare our algorithms with the previous version of our agent client to testify the improvements which contribute to our Apollo3D team's successes in wining various competitions at home and abroad.