| With the development of the manufacturing industry,the application of industrial robots is becoming increasingly widespread.In order to adapt to more complex production tasks,coupling visual servo technology with industrial robots has become a research hotspot in the field of industrial robots,compensating for the weak ability of industrial robots to independently collect information and enhancing their ability to perceive the environment.At present,in the field of light industry,visual servo robots are gradually applied to production processes such as logistics,freight handling,and mixed code sorting of packaging products.However,how visual technology can be better integrated with industrial robots and applied in actual production processes is still worth further research.Based on this background and application requirements,this study focuses on the following aspects:1.Build a robot platform based on visual servo control,with the KUKA KR10 R1420 robot as the main part of the motion trajectory,and use an RGB-D camera to collect color information,depth information,and point cloud information to achieve robot eye to hand target grasping.Simultaneously,the communication of the entire robot vision system architecture is completed through the ROS system,achieving the planning and output of high-quality robot paths,the output of packaging product related information,and the output of mixed code sorting related processes;2.Use the D-H parameter method to establish the forward kinematics model of the visual servo robot platform,and realize the linear mapping from the joint space of the robot to the Cartesian space.Based on the Monte Carlo method and the positive kinematics model,the theoretical model of the robot workspace is established,and the singularity model of the robot motion is established by linearly mapping the operability index to the robot workspace,revealing the anisotropic motion of the robot motion process Ability to provide an important theoretical basis for robot path planning;3.Under the service communication structure in the ROS system,an online inverse kinematics solver based on swarm intelligence optimization algorithm is developed,which realizes the low-latency and high-precision solution of real-time motion of general robots in inverse kinematics.In addition,a robot trajectory planning algorithm based on the combination of random sampling and polynomial interpolation is proposed,which can output the mixed-code sorting trajectory with low loss and smooth motion in real time;4.A 6D pose estimation network based on the combination of Transformer and CNN based on multi-modal data input is proposed,which realizes the high-quality extraction of color information and point cloud information features of packaging product images.A multi-level deep fusion mechanism is designed to achieve efficient fusion of multi-stream information.Through the regression analysis of the fusion features,the high-quality output of the attitude of multi-category packaging products is realized,and high-precision attitude information is provided for the theoretical calculation of the robot terminal kinematics;5.Based on the field research on the actual application scenarios of packaging products in printing and packaging factories and logistics industries,an experimental scenario of robot mixed-code sorting was designed,which realized different types and different postures in structural and non-structural working scenarios.The mixed-code sorting of packaging products verifies the stability and robustness of the robot system based on visual servoing in this study,as well as the superiority of the optimization algorithm.Through the above research,this paper realizes the hardware construction and software development of the robot system based on visual servoing,and completes the establishment,optimization and analysis of related theoretical models.This system can output mixed-code strategy information,target attitude information and obstacle-free trajectory information in the mixed-code sorting process of packaged products in real time,and has strong real-time and robustness.It can be seen that this research will provide powerful theoretical significance and application value for the development of industrial robot mixed code sorting system based on visual servo control. |
PDF Full Text Request