The Research On Semi-autonomous Control System Of Mechanical Arm Based On Machine Vision

In recent years,mobile robots equipped with a multi degree of freedom mechanical arm have been widely used in disaster rescue,space exploration,anti-terrorism explosion,defense science and technology and other fields,aiming to assist human beings to complete some basic operations.It is difficult for manipulators to manipulate themechanical arm.Therefore,the research of mechanical arm semi autonomous control based on machine vision is one of the hot topics in the mechanical arm of mobile robot.This paper mainly aims at the flexible manipulation of mechanical arm in the RoboCup Rescue Robot project,and completes the fouroperations of grasping,rotating,checking and touching the object.The RGB_D camera is used as the visual sensor,to study the problem of target location and recognition,motion planning,and feedback of the end pressure sensor,so as to realize the mechanical arm semi autonomous control system based on machine vision.Firstly,the P4P algorithm is used to determine the relative position and pose of the RGB_D camera in the world coordinate system,then,the target pose is estimated by the stable eigenvalue method,the vector method and the mean method.Then,the object is identified by the color histogram matching algorithm based on the texture feature.Forward and inverse kinematics analysis is carried out by D-H modeling based on the structure of the six degree of freedommechanical arm.Motion planning of the mechanical armare made in joint space and Cartesian space.The working space of the mechanical arm is analyzed,which determinesthe range of the working space of themechanical arm.And through the pressure sensor,which services as the feedback of the end of the mechanical hand,whether the object can be captured or rotated is determined.Based on the ROS platform,the control system experiment environment is built.Then,the modular multiprocess mechanical armlocal control and remote decision-making are written.The man-machine interface controlled by the manipulator is realized through the MoveIt! and Rviz tools.The manipulator can control the motion of the mechanical arm at the manipulative end.Finally,four kinds of operationstrategies are designed to grab,rotate,check and touch.The results show that the stability and rapidity of the robot's semi autonomous control can meet the requirements of the race well.