Research On Intelligent Sorting Robot System For Multi-target Stamping Parts Based On Machine Vision

Using molds to produce parts in batches has the advantages of high production efficiency,high consistency,low energy consumption,high precision and complexity,etc.,so it is widely used in machinery,electronics,automobiles,communications,aviation,aerospace and construction and other fields.However,in the continuous stamping production line with mixed layout of the entire sheet,due to the many types of molds and continuous stamping,the posture of the stamped multi-type metal parts cannot be fixed.This results in that traditional robots cannot complete the sorting of stamping parts in this scenario,and can only rely on manual sorting of stamping parts,which limits the production of stamping parts on a continuous stamping production line with mixed layout of the entire sheet material to a certain extent.efficiency.Therefore,on this basis,this paper integrates machine vision,neural network and robotics technology,and proposes a multi-objective intelligent sorting robot system for stamping parts based on machine vision.The intelligent sorting system introduces the intelligent identification technology combined with machine vision into the continuous stamping scene of the mixed layout of the whole sheet for the first time,and realizes the automatic functions of type identification,position return,positioning and grasping of stamping parts on the stamping production line.,get rid of the dependence on manual sorting in the continuous stamping scene of the mixed layout of the entire sheet,improve the production efficiency of stamping parts in this scenario,and provide a new solution for the sorting problem of stamping parts in this scenario.The subject has researched and analyzed the following aspects:(1)Research the calibration method of industrial cameras and how to use industrial cameras to collect images,and make data sets according to the selected labeling software;research the basic principles of neural networks,analyze some target detection algorithms and analyze Faster-RCNN target detection algorithms A detailed study is carried out,the constructed Faster-RCNN target detection algorithm is trained,and the target detection algorithm identification and localization tests are carried out.(2)Co-simulation of automatic grasping of stamping parts by robotic arm was carried out using python and the robot simulation software Coppelia Sim.In the process,the modeling method and communication protocol of the robot simulation software Coppelia Sim,as well as the kinematic modeling of the robot,the robot trajectory planning and other technical issues are studied.(3)A set of multi-target stamping parts sorting robotic arm grasping experimental platform was built.According to the stamping part type information and stamping part position information output by the neural network,the stamping part grasping experiment was carried out,and the reality of the multi-objective stamping parts intelligent sorting robot system based on machine vision was evaluated by the recognition accuracy and grasping success rate.feasibility.In summary,the multi-target stamping intelligent sorting robot system based on machine vision proposed in this topic can use Faster-RCNN target detection algorithm to identify stamping parts and output their position information according to the stamping image information collected by industrial cameras.,Based on these categories and position information,the sorting robotic arm can automatically perform kinematics calculation and path planning,and finally successfully grab stamping parts to complete the stamping part sorting task.To a certain extent,it can improve the sorting efficiency of stamping parts in the continuous stamping scene of mixed layout of the whole sheet.