Research On The Positioning And Guidance Method Of Robotic Arm Based On Target Recognition

With the improvement of socialist productivity and the rapid development of modern technology,robots have been applied to the detection and maintenance of transmission lines.Further research on more intelligent operation and maintenance technologies for transmission lines can ensure stable operation of the lines,effectively improve the intelligence and autonomy of robots,and free on-site maintenance workers from dangerous working environments.This article mainly studies the target recognition and positioning technology of the high-voltage transmission line shockproof hammer maintenance robot,including image recognition of shockproof hammer bolts,three-dimensional coordinate positioning of bolts,UR5 robotic arm coordinate system transformation,and path simulation of each joint of the robotic arm.The specific research content is as follows.Firstly,identify the shock absorbers and bolts in the transmission line.Due to the small size and difficulty in identifying bolts in the image,this article adopts a hierarchical detection method.Firstly,the SSD algorithm was used to identify the shockproof hammer and cut it.Then,the YOLOv4 algorithm was used to identify the bolts and eliminate interference from other bolts.Finally,the accuracy of the algorithm was tested.Secondly,calibrate the binocular camera.Select a binocular camera;Describing the relationship between the pixel coordinate system and the world coordinate system.Based on the operation process of the mechanical arm of the high-voltage line shockproof hammer maintenance robot,an eye out of hand calibration system was selected to complete the eye out of hand calibration of the mechanical arm.We selected the StereoCameraAccelerator toolbox in MATLAB to perform calibration experiments on a binocular camera.We obtained the internal and external parameters of the binocular camera through steps such as creating a calibration board,image acquisition,grayscale processing,and filtering out images with significant errors.Then,position the bolts to obtain the relative coordinates and distance between the bolts and the camera.Discuss the constraints and criteria to be followed in stereo matching,compare the actual and ideal models of the binocular vision system,and use the SGBM algorithm to perform stereo correction on the images collected by the binocular camera,obtain the image parallax of the left and right cameras,and match the corresponding points in the left and right cameras.Based on YOLOv4’s image recognition in the previous text,the distance between the bolt and the binocular camera,as well as the three-dimensional coordinates,were calculated using the parallax principle,and the accuracy and speed of the recognition algorithm were tested.Finally,the UR5 robotic arm was selected for simulation research.Describe the position and pose of the manipulator,select the UR5 manipulator to establish a coordinate system for each joint axis,build a D-H parameter model according to the parameters of the UR5 manipulator,and carry out forward and inverse kinematics analysis.On the basis of the above,the manipulator model is built in MATLAB using the toolbox to verify the accuracy of the manipulator kinematics operation.Then analyze the workspace of the robotic arm and eliminate singular points in the workspace.Finally,based on the bolt target position information identified earlier,a fifth degree polynomial is used for trajectory planning simulation of the robotic arm.