Research On Part Contour Extraction And Robot Motion Planning Based On Machine Vision

With the transformation and upgrading of manufacturing industry,intelligent manufacturing has become the main development direction.In intelligent manufacturing,robots are one of the core equipment.More and more production lines begin to use robots to replace humans to improve the level of automation.There are more and more application scenarios of human-computer interaction in intelligent manufacturing,but traditional industrial robots have low safety protection for people.Therefore,collaborative robots have gradually attracted people’s attention because of their high flexibility,high safety and easy operation,and they are increasingly appearing in various application scenarios.The traditional operating methods of robotic arm include manual teaching and offline programming.Manual teaching is inefficient in the face of small batch and diverse products,and it is not suitable for complex trajectory programming.Offline programming is suitable for products with small size difference between batches and complete 3D models,but it needs to calibrate the relevant coordinate system.In order to reduce manual operation and improve the applicability and efficiency of the robotic arm in different application scenarios,this paper studies the visual based robotic arm system,realizes the perception and response of the robotic arm to the external environment,and improves the flexibility and intelligence level of the robotic arm.The specific research content is as follows:Firstly,the equipment selection and construction of the robotic arm vision system were completed.The results of the point cloud acquisition test were analyzed to provide reference for the exposure time setting of the subsequent point cloud acquisition.In order to obtain the coordinate transformation relationship of the visual system,the camera calibration of the eye in hand is carried out.The methods of point cloud data preprocessing,contour point cloud segmentation and edge point cloud extraction are studied,and the edge point cloud optimization method is proposed,which can effectively improve the accuracy of edge point cloud and meet the point cloud quality requirements of robotic arm trajectory planning.Secondly,two kinds of robotic arm trajectory planning methods are proposed.One is planning method based on point cloud registration,which is used to solve the problem of difficult calibration and repeated calibration of offline trajectory.The other is based on edge point cloud planning method,which is used for edge trajectory planning of products with low consistency and no standard model.The variable strategy optimization nearest neighbor point cloud sorting method,a variety of trajectory attitude estimation methods and trajectory adjustment methods are proposed,which can meet the needs of complex edge point sorting and trajectory custom adjustment.In order to ensure the effectiveness of trajectory generation and prevent collision during the operation of the robotic arm,the kinematics model and interference model of the manipulator are established to realize the reachability check and interference check of the trajectory.Finally,in order to realize the above functions,the software system was developed,which mainly included four parts: user interaction interface,point cloud processing module,trajectory planning module and display module.Then,the developed system was used to test two trajectory planning methods.In the registration planning experiment,the offline trajectory could be applied to the photographed object,which verified the feasibility of the registration planning method.In the edge planning experiment,after the edge trajectory planning is completed,the trajectory is confirmed by using the developed simulation plug-in.Finally,the running results of the trajectory verify the effectiveness of the edge planning method.