Efficient reaching motion planning method for low-level autonomy of teleoperated humanoid robots

Abstract

This paper addresses an efficient reaching motion planning method tailored for teleoperated humanoid robots in complex environments. This method offers low-level autonomy that allows the robot to autonomously plan and execute simple tasks, thus making teleoperation easy. Efficiency is achieved by combining the phases of planning and execution. The planning phase quickly decides on a reaching motion by approximating mass distribution which enables analytical solutions of inverse kinematics. The execution phase executes the planned path while compensating for the approximation error, as long as other constraints are maintained. Simulations confirm that (1) a reaching motion is planned in approximately one second for the HRP-2 humanoid robot with 30 degrees of freedom in a constrained environment with pipes and (2) the execution is done in real-time.