Experimental kinematic identification and position control of a 3-DOF decoupled parallel robot

Abstract

This paper aims at using a kinematic identification procedure in order to enhance the control of a 3-DOF fully decoupled parallel robot, the so-called “Tripteron.” From a practical standpoint, manufacture errors lead to some kinematic uncertainties in the robot which cause real kinematic equations of the robot to be different from the theoretical ones. In this paper, using a white box identification procedure, the independence of degrees-of-freedom in the robot is studied. Considering the fact that the kinematic identification of a robotic manipulator requires the position of its end-effector to be known, in this paper “Kinect” sensor, which is a vision-infra red sensor, is utilized to obtain the spatial coordinates of the end-effector. In order to calibrate the Kinect, a novel approach which is based on a neuro-fuzzy algorithm, the so-called “LoLiMoT” algorithm, is used. Moreover, the results of experimentally performing the identification and calibrating approach are used to the end of implementing a closed-loop classic controller for path tracking purposes. Furthermore, the theoretical unidentified model was implemented in a sliding mode robust controller in order to compare the results with classic controller. The comparison reveals that classic controller which uses identified model leads to a better performance in terms of accuracy and control effort with respect to robust controller which is purely based on theoretical model.