Study On Key Technology Of Intelligent Recognition Of Workpiece Pose And Posi-Tion Parameters

With the advent of the "Industry 4.0" era,the demands for industrial modernization and intelligence are increasing day by day,and "intelligent manufacturing" has become the development direction of the industrial field.As the basis of production and manufacturing,Automated upgrade of industrial assembly lines are particularly important,and related technologies are becoming a research hotspot.Intelligent grab is the core link to realize the automation of industrial assembly line,and the systematic improvement and upgrading of related technologies cannot be delayed.Workpiece recognition is the key to intelligent grab.Therefore,researching on the intelligent recognition technology of workpieces is very important.This topic is a sub-project of the “Intelligent Industrial Robot Sensing and Execution Technology Research” of the Key Science and Technology Development Project of the Science and Technology Department of Jilin Province.It mainly studies the problem of intelligent recognition of mechanical workpieces during the process of industrial robots intelligently grabbing mechanical parts on the automated industrial assembly line.This paper proposes a method for intelligent recognition of workpiece pose based on workpiece 3D laser point cloud data,which can identify the type and pose of the workpiece.First,the specific mechanical workpiece used for the experiment was scanned to obtain its Three-dimensional laser point cloud data by the measurement system of the Two-dimensional laser galvanometer scanner.Second,identify the type of artifact.The workpiece point cloud is center-divided,and the workpiece point cloud slices in the three coordinate axes are obtained respectively with the MATLAB software.The HALCON software is used to denoise,enhance,and segment the point cloud slices,and extract the outline information of the point cloud slice.Finally,according to the characteristic geometric parameters of the slice profile,combined with the mechanical three-view theory,complete the judgment and identification of the workpiece type.Then,using 3D point cloud registration technology,the pose recognition process is divided into initial recognition and accurate recognition.In the initial recognition,the appropriate threshold is selected according to the variation law of the point cloud data normal vector,and the feature point set for the initial recognition of the workpiece pose is screened out.Afterwards,the improved random sample consensus(RANSAC)algorithm is used to roughly match the target point cloud and the template point cloud to complete the initial recognition of the workpieces pose.In the stage of accurately identifying the workpiece attitude,the iterative closest point(ICP)algorithm was improved,and the improved algorithm was used to complete the accurate registration of the point cloud.The unit quaternion method is used to solve the initial recognition and accurate recognition results,and the attitude parameters are obtained to complete the posture recognition of the workpiece.Finally,in order to verify the feasibility and effectiveness of the workpiece pose intelligent recognition scheme,large number of simulation comparison experiments were carried out on the workpiece recognition and pose recognition methods.The experimental results verify that the intelligent identification scheme for mechanical workpieces on industrial assembly line is effective,and the design scheme of this project has great reference significance for solving similar problems.