The work aims to propose an end-effector positioning method based on machine vision, so as to improve the positioning accuracy of the package handling robot. Based on OpenCV, a visual calibration algorithm was designed, which included camera calibration and pose calibration. The transformation between image coordinate of the object to be stacked and manipulator coordinate was realized. The control system was designed based on the IPC and the motion control card. In the meantime, design methods of hardware and software were given. Finally, experimental studies were carried out, including origin positioning and repeated positioning. The experimental results showed that, the proposed control method could improve the positioning accuracy of the handling robot, with the origin positioning error of about 0.14 mm and the repeated positioning error of about 0.6 mm. The positioning method of the handling robot can meet the requirements of packaging and palletizing.