Research Of Position-based Visual Servoing Control System Of Industrial Robot

Machine vision is being widely used recently, putting this technology into industrial robot control system will obviously increase its intelligent and automation level, and broaden the application field of robot. Compared with the traditional machine vision, visual servoing is more flexible and accurate, more robust to calibration error of robot. So, based with ABB-IRB120 robot, CCD camera, conveyor, this paper construct an industrial robot visual servo control system, and discussed several problems about the system. The main research works are as follows:This paper discussed and analyzed the theory of robot visual servo control, summarized the possible development directions of this technology and its existing problems of present stage, According to the specific requirements, this paper accomplished the hardware selection, system layout and software process design.The image processing speed and accuracy are important factors affecting the performance of the robot visual servo system. To solve the problem, this paper analyzed The traditional Grayscale –based image matching method and Edge-based matching method in the processing speed and matching accuracy. For the traditional NCC-based matching method with the time-consuming problem, this paper propose an improved image pyramid hierarchical search strategy which improved the algorithm matching speed.This paper discussed the common camera calibration method in robot visual servo control. With the Eye-in-hand configuration, this paper presents a modified constant rotation matrix method to estimate the relative pose between robot and target. The calibration precision was verified by experiments.This paper deeply analyzed the PBVS control mode. This paper constructed the kinematic model of ABB-IRB120 robot with its link parameters. With Robotics Toolbox in Matlab, this paper established mathematical model of the robot visual servo system and completed the simulation of PBVS system in Simulink.Taking the delay of the visual system and servo control algorithm into account,To make the visual servo system more real-time, this paper used Kalman filter processing the visual feedback information. The experiment results showed that, Kalman filter compensated the tracking lag caused by visual delay effectively. It improved the respond speed of the system significantly without affecting precision.This paper finally construct the robot visual servo control system for target locating and tracking. Experiment results verified the method and conclusion present by this paper.