Deep Learning Based Delta Robot Stud Welding Extension Rod Sorting System Research

With the advent of the era of Industry 4.0 intelligent manufacturing,more and more production lines introduce industrial robots to achieve the flexibility and intelligence of the production line,and the sorting system as an important part of the production line,its intelligent research is of great significance.Delta parallel robot has the advantages of high rigidity,fast speed,strong flexibility,light weight,small size and low cost,and has unique advantages in sorting lightweight,multi-type and multi-size parts.The deep learning model can quickly build a flexible visual recognition system with high flexibility,high stability,high adaptability and low cost.The study of Delta parallel robot sorting technology based on deep learning to realize a set of intelligent extended rod sorting system with high flexibility,high speed,high stability and low cost is in line with the development trend of intelligent industrial assembly line,and has important theoretical and practical significance for realizing intelligent industrial production.The main research contents are as follows:1.Research on deep learning-based target recognition models.Develop and implement Squeeze network target recognition algorithm,construct visual measurement model,adopt Zhang’s calibration method to achieve camera calibration,and implement deep learning based grasp detection algorithm.2.To study the Delta-type parallel robot mechanism.The kinematic model of the robot’s mechanical structure is created by analysing the forward and inverse solutions of the position,and the robot mechanism workspace is obtained on the basis of the analysis of the positive position solution,which provides the theoretical basis for building the mechanical structure design of the experimental prototype.3.Research on the Delta parallel robot control system.Based on STM32,the underlying numerical control layer system is constructed,and the parallel robot control algorithm is developed to realize the motion trajectory control.Build the host computer system based on the PC,develop the communication protocol of the underlying system,and build the system software UI interface.4.On the basis of the above theoretical research,the Delta-type parallel robot for stud welding extension rod sorting intelligent vision sorting experimental prototype is developed to achieve accurate recognition grasping classification,to verify the correctness of the theoretical analysis of the parallel mechanism in this paper and the feasibility of building a CNC system.