Vision-based Computed Torque Control for Parallel Robots

Abstract

This paper presents a novel approach to dynamic control of parallel robots based on the end-effector pose and velocity visual measurement. We show that computed torque control in joint space is unsuitable for parallel robots and that computed torque control in Cartesian space requires the end-effector pose and velocity estimation. Relying on a recent work showing ability of camera to measure both the pose and the velocity of a mobile object, we propose an original vision-based computed torque control scheme. Improvements are discussed and simulation results show the expected performances