A Novel Online Whole-Body Motion Planning Algorithm for Supervisory Control of a Legged Mobile Manipulator

Abstract

In this paper, a novel online whole-body motion planning algorithm for supervisory control of a legged mobile manipulator (LMM) is presented. The inputs to the proposed algorithm are the updates to the end-effector’s desired position in Cartesian coordinate system, given continuously at a constant frequency. The algorithm plans the trajectories and calculates the joint space parameters of the LMM in real time and returns all the joint angles continuously, at the same frequency, as the outputs. In the current implementation, the incremental inputs are given to the algorithm using a joystick by the user. To ensure a singularity-free motion of the LMM, the proposed algorithm plans adaptive stroke lengths for the trunk body. The algorithm is designed to be highly modular, so that any necessary improvement to it can be made easily. The proposed algorithm is tested by implementing it on a virtual robot in Gazebo simulation and then, validated on a physical prototype of a hexapod mobile manipulator.