The Techonology Research Of Robot Grasping Based On Visual Servo

Because of the widespread use of robot and the study of robotics,the most basic and most important ability of robot-grasping has been studied continuously.Robotic grasping technology has been mature and widely used in the production of industrial automation,providing a lot of convenience for industrial automation production.However,the family environment is different from the industrial production environment,it is unstructured and the placement of objects is uncertain,or objects block each other,and so on.So,how to use the robot in the home environment is a problem that is difficult to solve effectively.In order to solve the autonomic grasping of the service robot in the home environment,this paper uses the depth camera as the visual sensor and uses the robot arm to build the robot hand-eye system,and builds the intelligent grabbing software platform of the robot grasping.The main content and research results were as follows:[1]Kinematics mathematical modelling and solving for the robotic experimental platform to ensure the precise movement of the end effector in the robot gripping process;[2]Realsense is used to collect environmental depth information and use PCL to convert it into point cloud information,and preprocess the point cloud information to separate objects and store them separately for processing.[3]For the problem that traditional methods cannot effectively identify and locate transparent objects,the use of depth information and refraction of light to identify and locate transparent objects has been proposed.This method can effectively distinguish whether objects are transparent and obtain three-dimensional model of objects;[4]Aiming at the time-consuming or need to collect a large number of samples in the existing object-grabbing methods,a grasping method using depth information is proposed,and a grabbing contact line model is proposed,on the gradient profile of the object-depth image.Finding the best place to grab[5]The robot's autonomous grasping software system is built,and cross-compilation is performed using Matlab and C++.Each function is packaged as a functional module.The system is used in the actual scene for testing to verify the feasibility of the algorithm.