Research On Position And Attitude Measurement Method Of Automotive Sheet Metal Parts Based On Machine Vision

In the automotive sheet metal workshop,most of the automotive sheet metal parts need to go through complex processes such as shearing,welding,riveting,bending and snapping,bending and forming,etc.During this period,the workpiece needs to be transferred between different processing stations,which leads to a lot of time and manpower to repeat the process of loading and unloading the workpiece,resulting in low workshop productivity and rising costs.At present,although industrial robots have been gradually used to replace manual loading and unloading tasks,most of them are still through the robot "teaching" method to guide the robot to grasp the workpiece.This method can only follow the prescribed path of action,no independent identification and positioning capabilities,the level of automation needs to be improved.In order to solve the above problems,a low-cost,high-precision,intelligent and practical workpiece position measurement system based on machine vision is designed to assist robots in achieving accurate and fast loading and unloading,which can improve the intelligent level of robots and control costs.The main contents of this paper are as follows:(1)The system hardware and software selection,the construction of the workpiece 3D point cloud acquisition platform are completed.The principle of the camera model and the conversion relationship between coordinates are studied.The camera calibration is completed and its reconstruction accuracy is verified.Through the study of the point cloud filtering algorithm and the streamlining algorithm,the field point cloud acquired by the camera is realized to remove noise and reduce the number of points without losing features.(2)This paper completed the task of segmentation and identification of target artifacts in the field attraction cloud.To improve the efficiency and accuracy of the subsequent alignment algorithm,the artifact is segmented from the field point cloud using a segmentation algorithm.In this paper,I used the template matching method to complete the identification task,which identifies the target artifacts from the scene by building an offline template library and using a point cloud recognition algorithm.(3)I have completed the study of 3D point cloud pose estimation techniques for automotive sheet metal parts.Through the study of various point cloud alignment algorithms,the alignment scheme was determined,which is to obtain the initial poses by coarse alignment first,and then obtain the exact poses by fine alignment.(4)In this study,I have designed and completed experiments on the positional measurement of automotive sheet metal parts.Repeated experiments were carried out on the experimental platform,which proved that the accuracy and real-time performance of the method met the requirements.In this paper,I have proposed a new method of using a depth camera to achieve positional measurement of sheet metal parts based on the actual situation in an automotive sheet metal shop.This method used the Ensenso N10-304-18 camera to acquire a three-dimensional point cloud of the workpiece.After the point cloud is filtered and refined,the workpiece point cloud is segmented by using the improved RANSAC algorithm and the Euclidean segmentation method.Next,I have completed the recognition task for the parts by using the template matching method for object recognition based on corresponding point classification.Experimentally obtained recognition success rate of 71.7% in the case of occlusion and 96.7%in the case of no occlusion.Finally,I have obtained the target workpiece pose by combining the coarse alignment algorithm of Super4 PCS with the improved ICP exact alignment algorithm.The experimental results with and without occlusion achieved position error Δd <4 mm,angle error Δθ < 4.5°,and calculation time t < 6 s.The results have proved that the method can meet the actual production needs of workpieces.Therefore,this method is of great significance to reduce the continuously rising labor cost and accelerate the complete automation of automobile production line...