Machine Vision Based Location And Attitude Estimation For Circular Workpiece

With the pressure of reducing labor resources and the development of technologies such as computer technology and image processing,industrial robots with machine vision aided control have been widely used in the field of manufacturing.In many existing applications,machine vision is often used to locate objects in the plane to guide the industrial robot for processing operations.In order to further enhance the intelligent level of industrial robots,to measure the position and posture of the target in three-dimensional space with computer vision has become a new hot spot in the field of machine vision and robotics.In the field of industrial production,many objects processed by industrial robots have circular features and no obvious texture,so it is difficult to estimate the position and pose of such objects by extracting feature points.In this paper,based on the machine vision technology,ellipse detection technology,the binocular location technology,it is researched to estimate the location and posture of a space target with a circular feature of known radius,and the specific application scenario is the automation assembly of umbrella cover of large sunshade.The main work of this paper is as follows.The camera imaging model was established based on the imaging principle of the camera,considering the influence of distorting factors in the imaging process.And the calibration of the camera was carried out by using the camera calibration kit of Matlab to get the internal and external parameters of the binocular camera.Use Canny edge detection algorithm to process the image to get the binary edge image,and then use the Hough detection algorithm based on the midpoint of the chord to coarsely locate the midpoint of the ellipse.And based on this,use a stochastic ellipse fitting algorithm based on least squares to detect and fit the target ellipse.In practical applications,do not need to rotate the camera.Construct an elliptical observation vertebral model,keep the optical center of the camera constant,rotate the ellipse to a specific position,and construct a hypothetical ellipse perspective projection.Through the geometric properties of the projection,the pose of the space target plane in the camera coordinate system after rotation can be estimated and then it can be converted into the actual camera coordinate system.Finally,with the help of binocular vision,the center of the circular target can be obtained,and the ambiguous problem of solving the circular target pose can also be solved.Finally,the machine vision industrial robot experimental system was built up,and the scheme described in this paper was used to realize the positioning and pose estimation of objects with circular features in three-dimensional spaceoSpecific to the test,the position and pose of the parasol umbrella cover can be estimated,and then the industrial robot is guided to clip and screw the umbrella cover.