With the development of science and technology, more and more robots are applied to industrial production. But the traditional industrial robot can only move according to the planned trajectory in advance, it can’t obtain outside information, which limits the application range. With the development of machine vision in recent years, the robot control technology based on machine vision is developed. But a lot of machine vision applications are monocular camera, and is fixed in a certain position in the transmission belt, the machine vision system of this form can’t locate the object accurately.In this paper, the technology of machine vision and robotic arm are discussed and research. The combination of binocular vision system and robot arm can achieve the recognition, location and grasp of the target. Firstly, the binocular vision is calibrated to determine the parameters of the camera. And then the "eye-in-hand" camera calibration, to determine the relationship between the end effector of robot coordinate system and camera coordinate system.According to the characteristics of the target, the target is identified. The different color of the rings are as the target to be detected in experiment. The ring is extracted after a series of image processing, such as color recognition, ellipse fitting, and so on. Ring positions in images can be roughly located by the color recognition, then the region of interest (ROI) is the part of the ring position. The contours can be extracted in ROI of image and using ellipse fitting for contours, then the center position of the ring is determined. Do the above processing on both left image and right image, respectively, to get the center of ring in the left and right images, and the three-dimensional coordinates of the ring relative to the camera can be obtained by using the triangulation method. Finally according to the "eye-in-hand" calibration results, the ring’s coordinate can be transformed into the coordinates of the robot. |