Dynamic path tracking of industrial robots with high accuracy by visual servoing

Abstract

In this paper, a practical dynamic path tracking (DPT) scheme of a 6-DOF industrial robot is presented. The DPT scheme is a position-based visual servoing (PBVS) to realize 3D path tracking by correcting the robot's movement in real time with eye-to-hand visual measurement feedback. In the traditional task-implementation mode for industrial robots, the task planning and implementation are taught manually. The proposed DPT scheme can realize automatic pre-planned task and improve the tracking accuracy. Moreover, a Kalman filter is proposed to obtain smooth pose estimation and reduce the influence caused by image noise, vibration and other uncertain disturbances. The developed DPT scheme can be integrated with the industrial robot's controller without accessing the robot joint control or retrofitting with high-end encoder. By using C-track 780 from Creaform, the experiment on Fanuc M20-iA with the developed DPT scheme demonstrates the improved accuracy of the robot compared with the traditional robot controller.