Robust position-based visual servoing of industrial robots using feedforward kinematic approach based on integral quasi-super twisting algorithm

Abstract

This paper presents a novel robust kinematic control approach based on feedforward inverse kinematic compensation for real-time pose correction of vision-based industrial robots to enhance the trajectory tracking accuracy. The conventional methods entail the robust design of the dynamic controller to handle the uncertainties in the dynamical model, which is not applicable to most industrial robots since they operate based on their built-in controller designed by the manufacturer and usually are not accessible. The proposed method, however, is a robust kinematic controller capable of handling the uncertainties in both dynamic and kinematic models. To this end, first, the robot's pose is estimated by a nonphysical contact sensor (in this research, a photogrammetry sensor). Then it is fed to the kinematic controller for the real-time pose control task. The feedforward part of the proposed controller, which is the inverse kinematic function of the robot, is considered in the control system design to reach the highest possible accuracy, especially for trajectory tracking purposes. The proposed method utilizes a novel integral quasi-super twisting algorithm (IQSTA) as the compensator within the control loop to reach a finite-time convergence with very high precision and robust performance with minimal chattering. The stability analysis of the proposed method is presented. The experimental results on an industrial robot (Denso VP-6242) equipped with an ATEMEK photogrammetry sensor show the superiority of the proposed method over other state-of-the-art approaches.