The Application Of Image Processing And Multi-sensor Information Fusion In Pipeline Intersecting Wire Weld Inspection Robots

In the field of non-destructive testing,the detection of intersecting line weld is difficult and the detection efficiency is low.The intersecting weld detection robot,adsorbed on the wall of the pipe by the magnetic wheels,needs to move along the intersecting line weld and detect weld defects carrying non-destructive testing equipment.The robot also needs to constantly adjust the angle and distance between the probe and the weld during the detection process.Therefore,the automatic non-destructive detection robots,which aimed at the detection of intersecting line weld,must have the ability to automatically track and locate weld defects.Based on the detailed analysis and consideration of the actual demand and field application scenarios for non-destructive testing of intersecting line weld,and contrast to a variety of robotics commonly used in the field of automatic tracking and locating scheme,this paper establishes the scheme of visual guidance through the automatic identification of the magnetic color-coded strips and accurate locating through the multi-sensor information fusion.Aiming at the established scheme of visual guidance and information fusion,an image processing algorithm based on machine vision guidance scheme is proposed in this paper.Based on the color space conversion and color edge detection,an image pyramids was first introduced to the Hough transform line detection,to identify the magnetic color-coded strips quickly and achieve the real-time control of the attitude and motion trajectory for the robot;a multi-threshold calculation was further applied to the region growing algorithm to segment the artificial mark points,and then the artificial mark points were counted to assist robot to achieve positioning.Then,the robot's motion trajectory and its own attitude are modeled and deriving the mathematical relationship between the attitude and the motion trajectory of the robot in the course of movement,according to the restrictive condition that the robot must move along the intersecting line weld.The odometer locating model is established by using the photoelectric encoder and the counting of the artificial mark points,and the IMU locating model is established by IMU system.According to the derived mathematical relationship,Kalman filter is used to fuse the information of IMU model and odometer model to achieve the robot localization function based on multi-sensor information fusion.Through the MATLAB software,the odometer model,the IMU model and the multi-sensor information fusion model are simulated,and the results and errors of the three locating models are analyzed and compared.Experiments show that the proposed image processing algorithm can make the robot stably identify the magnetic color-coded strips,so that achieve the automatic tracking function of the robot.At the same time,the image processing algorithm can accurately segment the artificial mark points to assist the robot to achieve locating function.The simulation results of the computer show that the multi-sensor information fusion based on Kalman filtering can achieve the robot accurately locate itself,and its locating accuracy is significantly better than using the photoelectric encoder or IMU system alone.