Research On Recognition And Orientation Technology Of Scattered Workpieces Based On Binocular Vision

With the increasing intelligence of industrial development,the application of industrial robots in industrial manufacturing will also become one of the mainstream trends.The industrial site often has positioning and grabbing operations for scattered workpieces.Due to the tilting,stacking,and blocking of the scattered workpieces,the more common monocular vision technology is difficult to solve its identification,orientation and grasping problems.At present,the scattered workpieces are generally placed flatly by manual or special mechanisms,and then monocular vision is used for subsequent operations,which seriously affects production efficiency and increases manufacturing costs.Therefore,in order to improve the production efficiency of the factory and promote the further development of robot automation,this paper conducts research on the identification and orientation technology of scattered workpieces based on binocular vision for the orientation and grabbing of scattered workpieces.The main research contents are as follows:Firstly,analyzing the task requirements of the subject,summarizing the existing technical difficulties,and providing an overall solution;completing the hardware selection,and finally determining the parallel binocular camera installation scheme used in this article,and give specific parameters;then analyzing the model of the camera imaging,establishes the coordinate system of the vision system,and combines the distortion model to derive the transformation relationship of the projection matrix.The Zhang Zhengyou calibration method is used to calibrate the vision system constructed in this paper.Secondly,pre-process the acquired images according to the actual working conditions,and then based on the analysis of the traditional template matching method based on SIFT descriptors,the target postures of the actual target workpiece due to tilt,stacking,and occlusion change and The problem of incomplete features and the shortcomings of the original algorithm itself are poor in real-time.Two optimization schemes are proposed and the performance of the two algorithms is evaluated through experiments.The results show that the optimization method based on the weight threshold has a better match effect.Thirdly,for the problem of high false matching rate in the occlusion area,an optimization scheme is proposed,and the improved algorithm has better matching effect and real-time performance through experiments;Then,the edge of the identified graspable workpiece is fitted,and the three-dimensional reconstruction of the fitted edge of the workpiece is performed using the parallax value and the calibration result obtained by the stereo matching;and the spatial pose of the scattered workpiece is obtained by the spatial circle fitting of the spatial point Information,and put forward the evaluation standard of the grab strategy based on the inclination of the workpiece surface and the imaging plane.Finally,an experimental system for the identification and positioning of scattered workpieces based on binocular vision was built.The Eye-to-Hand mode was used to perform hand-eye calibration on the system,and the spatial and posture information of the scattered workpieces in the camera coordinate system was converted to robot's base coordinate system.Verifing the orientation accuracy of the binocular vision system and the positioning grab experiment for scattered workpieces.The results show that the system in this paper can meet the requirements of identification and positioning of scattered workpieces,and the experimental results are good.