Research On Visual Navigation Techniques For Snake-arm Robot

Machine vision is an important approach that enables robot to have perceptual ability just like human being.The robot can be more intelligent after integrating visual information into robot system.As a kind of special robot,the snake arm robot is mainly used under a complex non-structural environment,and the machine vision is an important guarantee for the smooth operation of the snake arm robot in such an environment.Visual positioning navigation,also known as visual servo,which use image information collected by robot image sensor as feedback signal of close-loop control of robot.Snake arm robot arm has multiple degrees of freedom,which determines that its working environment is a proper 3D space,therefore the visual navigation techniques of snake arm robot are different from traditional ground mobile robot.This study presents an image-based visual servo system of wire-drive orthogonal-joint snake arm robot,which realize the spatial visual navigation of snake arm robot.The main work is as follows:First,the mechanical structure of wire-drive orthogonal-joint snake arm robot is introduced,including the orthogonal joint structure,the lever drive mode,the driver installation,etc.,and the design parameters are explained.The model of the snakelike robot arm is constructed using D-H parametric method,and the relevant kinematical equations is deduced,based on the kinematical equations the position relationship between the robot arm end and base coordinate system is obtained.Moreover,we design the control system of the snake arm robot,and the schematic diagram of the related hardware circuit is given,finally an effective control test platform for snake arm robot is set up.Secondly,this paper introduces common image processing methods in machine vision,including image enhancement and smoothing,image characteristics extraction,image segmentation based on threshold value and feature-based target identification method.The algorithm is implemented in Open CV,by comparing the actual treatment effect,choice the suitable algorithm for the visual navigation control of the snake arm robot.Subsequently,we derive camera's pinhole imaging model and obtained the transfer matrix among world coordinate system,camera coordinate system,image coordinate system and pixel coordinate system.According to the basic principle of zhang zhengyou calibration method,the camera parameters solution formula is deduced,and successfully obtained the camera parameters and the distortion parameters of the camera,on the basis of those parameters,we correct the image of this camera.Finally,this study introduce relevant concepts of visual servo,deduce the objective function for uncalibrated visual control.The image Jacobian matrix is deduced on the basis of camera imaging model,and uncalibrated visual servo controller is designed according to the image Jacobian matrix.And design the target tracking algorithm based on SIFT and Camshift.Then carry out the target tracking system for snake arm robot,and the design of the target tracking algorithm and controller is verified on the system.Experiments show that the designed platform can realize the distributed control of the serpentine robotic arm system.The designed control method can achieve the visual servo control of the serpentine robotic arm,which has faster response speed,higher positioning accuracy and repeatability..