The Research Of Assembly Robot Based On Visual Shaft Hole Identification And Positioning Technology

Recognition and positioning is an essential step in the assembly process,and it also act as an important symbol to measure the ability and level of the robot's perception and intelligence.In this paper,the vision system of the robot assembly is combined with the robot technology,and the assembly of the robot shaft axis is taken as the object.The matching of the shaft hole and the target location technology are deeply studied.The main work of this paper is as follows:First of all,the background and significance of this topic is introduced,and this paper summarizes the development and current situation of machine vision technology at home and abroad.The target recognition and localization technology which is designed in this paper is analyzed in detail.Secondly,based on NACHI six degree-of-freedom robot,the robot kinematics foundation is introduced,and the robot D-H model is established.By analyzing and deriving the robot positive kinematics equation and inverse kinematics equation,the technical support for the robot's target location is provided.For the robot vision system,which is classified by different constraints,the camera imaging model,camera calibration significance and calibration method are introduced.Thirdly,pictures taken by the camera are preprocessed,including the linear change of the image gray level and the histogram equalization,furthermore,this paper introduces several commonly used methods of filtering noise reductio.The peak signal to noise ratio(PSNR)and the mean square error(NMSE)are used as the criterion to determine the noise reduction method of Gaussian filter with good smoothing effect.For the two value processing of the image,the threshold of the image segmentation is determined by the maximum inter class variance method(OTSU).The simulation results show that the method can effectively realize the segmentation of the target workpiece and the background.Then,the extraction of the workpiece edge features,several common edge detection operator analysis and comparison are introduced to determine the use of Canny edge operator to extract the edge of the workpiece contour information.Fourthly,the target contour matching is studied,and for describing the contour features of the target,the geometrical moment invariants of Hu are introduced to match the target contour.In order to calculating the target axis pose,the workpiece axis shape is determined by the minimum circumscribed moment method and the Hough linear transformation Heart and bus.For the hole circle detection,a Hough transform circle detection algorithm based on improved gradient information is proposed to detect the hole circle.The invalid accumulation of the standard Hough transform in constructing the accumulating array is reduced by using the gradient information of the target edge point,and the operation efficiency of the algorithm is improved.Fifthly,the positioning of the axis of the mounting robot introduces the vision positioning algorithm based on monocular.By moving the end of the robot to control the movement of the camera and using the camera to photograph the target in different pose to extract the matching feature points.Through the calibration of the eye of the EIT system,the transformation matrix of the relative end effector of the camera are obtained,and it also obtain the rotation matrix R and the translation vector T.Finally,the extraction and measurement of axial hole features in the image coordinate system are simulated by MATLAB,and the three-dimensional coordinates of the midpoint in space are obtained to achieve the target positioning.In the end,the paper summarizes the research contents,points out the shortcomings in this study and prospects the next work in the future.