Design And Implementation Of Hand-eye Robot Control Software Platform Based On ROS

EAST Tokamak’s core device will generate a lot of radiation and super electromagnetic fields during operation,so it cannot be maintained manually at close range.In order to meet the remote handling maintenance requirements of the tokamak device,it is necessary to construct a hand-eye robot system with a visual sensor at the end of the robot arm,enter the vacuum chamber to complete the observation task,and replace the damaged graphite tile on the inner wall.This involves the accuracy of robotic arm control and vision sensors.Robot application development and robotic arm control are the most important content in the hand-eye robot system,and also the basis for moving the hand-eye robot.At the same time,due to insufficient light inside the core and distortion of the imaging of the visual sensor,the accuracy of the image is reduced,which will further affect the accuracy of the posture control of the robotic arm.This thesis takes the UR5 robotic arm and Xtion Pro Live visual sensor as experimental objects to study the problems of hand-eye robot system construction,realizes the development of robotic arm control software and the calibration of visual sensors,and completes the hand-eye calibration experiment on this basis.The main research contents are as follows:(1)Positive kinematics and inverse kinematics analysis and simulation of UR5 robot arm.For the control process of the robot,not only the spatial posture matrix of the end effector of the robot to be solved using positive kinematics,but also the joint angles of the robot under the current posture using inverse kinematics.Based on the established D-H parameter table of the UR5 robot arm,the kinematic analysis of the UR5 robot arm is completed,and the relationship between each joint angle and the position and posture of the end effector is obtained.At the same time,based on the solution obtained by the positive kinematics,the UR5 robot arm is completed using an analytical method The inverse kinematics analysis of the system provides a foundation for the acquisition of experimental data for hand-eye calibration.Finally,the correctness of the kinematics model of the UR5 robot arm is verified by the V-REP simulation environment.(2)Analysis and modeling of Xtion Pro Live vision sensor imaging and Eye-in-Hand hand-eye system.In practice,there are certain errors in the assembly and processing of the optical components of the camera,which results in a non-linearrelationship between the size and shape of the image acquired by the camera and the actual image,that is,graphic distortion.For the camera after graphic registration,it also needs to pass the hand-eye calibration conversion to obtain the conversion relationship between the camera and the end of the robot arm,and then the robot can be accurately controlled by the image.To solve the above problems,the camera imaging model was established based on the Xtion Pro Live visual sensor,image distortion analysis was performed,and the internal reference calibration of the camera was completed using the Zhang’s calibration method;based on the conversion relationship of the hand-eye system coordinate system based on Eye-in-Hand structure,the hand-eye calibration equation was proposed,Using a two-step method to find the coordinate conversion matrix,to provide a theoretical basis for the following hand-eye calibration experiment.(3)Design and implementation of application development tools for ROS-based hand-eye robots,including Ros Launch visual creation,function module graph(View Graph),project deployment,robot arm control,and robot status feedback.At present,the threshold for robot application development is relatively high,which is not conducive to the promotion of robot applications.This paper proposes a hand-eye robot software platform based on a plug-in form,modularizes the robot functions,completes the development and assembly of the hand-eye system project using visual development methods,and implements the establishment of a hand-eye calibration experiment software environment.(4)Based on the hand-eye system composed of UR5 and Xtion Pro Live,we test the function of the hand-eye robot control software platform.Using this software platform to complete the robot arm control and image data collection,after the hand-eye calibration calculation,the posture error of the hand-eye system constructed in this paper is only 0.03,and the position error is only 0.0469 mm.Experiments show that the robot control platform designed and developed in this paper can accurately and quickly complete the control function of the robot arm.