Sim-to-Real Hierarchical Planning and Control System for Six-Legged Robot

Yue Gao, Yangqing Fu, Ming Sun

Abstract: Legged robots have attracted much attention both from industry and academia. Despite the recent progresses in robotics, planning and control in complex environments are still great challenges for legged robots. Generally, constructing the planning and control system for legged robots requires complex designing and parameter tuning. To reduce resource consumption and potential risk in this process, a sim-to-real hierarchical planning and control system is proposed in this paper. The proposed hierarchical system can improve the data efficiency and reduce the training cost utilizing the reinforcement-learning-based framework. Several experiments are conducted to demonstrate the feasibility and effectiveness of the proposed method.