Whole-body Motion Planning and Control for Underactuated Wheeled-bipdal Robots

Abstract

In previous works, the motion of wheeled-bipedal robots is divided into balance and movement of wheels and locomotion of legs. However, the two parts have strong coupling and huge disturbance with each other. In this paper, a whole-body motion planning method is proposed, which can achieve the control tasks of torso pose and full-body dynamic balances simultaneously. According to the dynamic analysis of the wheeled inverted pendulum model (WIPM), the horizontal posture is introduced into the torso pose to establish the kinematics with the torso pose as the task space. Combined with the kinematics of the single wheeled-leg, the whole body kinematics model is obtained, and the whole-body Jacobian (WBJ) matrix is derived to establish the speed relationship between the task space and the joint space. Based on the above motion planning, we designed a task space controller and a joint space controller. In the virtual environment, wheeled-bipedal robot is controlled to velocity tracking and yaw rate tracking when turning, which proves the effectiveness of this motion planning method.