The Robot Arm Grab Simulation Study Based On Posture Estimation

With the development of society and the shortage of human resources,under the urgent need to improve production efficiency,intelligent machines have been praised by people.Driven by this trend,intelligent industrial robots equipped with visual sensors have obtained extensive promotion and application.This article focuses on object pose estimation and virtual simulation platform grasping for research,in the early stage of object pose estimation,the camera parameters used in the project are explained,and internal parameter calibration experiments are carried out,in terms of pose estimation,it is based on LINE-MOD objects.The recognition algorithm is improved.According to the type and characteristics of the object,the corresponding template production method is selected,and different pose estimation strategies are carried out for two-dimensional and three-dimensional objects.In the three-dimensional object pose estimation method,according to the different shape features to be recognized,analyze the arguments from the three aspects of distance similarity ratio,depth averaging,and point cloud matching,in the virtual simulation capture stage,the pose information is obtained from the pose estimation stage of the object,and the pose information is imported into the ROS system through the topic form,and the Move It control platform is used.Path planning and motion control of the virtual manipulator,after the camera's external parameter calibration,the coordinates of the object in the base of the manipulator can be obtained,and then Move It is used to control the manipulator to simulate and capture the recognized three-dimensional objects.In this paper,the process design and experiments are carried out from the object pose estimation to the virtual simulation capture,through the extraction of the LINE-MOD algorithm,the algorithm framework is made more concise,and the object template is designed according to the shape and texture characteristics of the object.Method,for objects with symmetry in the shape characteristics,not only the number of corresponding templates is reduced,but also the production time of the templates,for the acquisition of distance information in the pose estimation process,diversified methods are used for different objects,the purpose is Applicable to various types of objects,and apply the algorithm in this article to the pose estimation of various objects more quickly and accurately,for the grasping process of objects,this article is based on ROS and uses Move It to control and generate control information.For many on the market Both the robotic arm and ROS already have a hardware interface control package.With a simple configuration,the control information can be transmitted to the robotic arm control cabinet through the existing robotic arm configuration file,so that the real robotic arm can be the same as the visualized robotic arm in rviz.This process simplifies the robot arm development process,shortens the deployment time of the robot arm,and is of great significance to the study of industrial robots equipped with identification systems for the "machine generation".