Research On Accurate Positioning Technology Of Mobile Robot In Coal Mine Based On Multi-sensor Data Fusion

The Coal Supervision Bureau proposed the "Key R&D Catalog of Coal Mine Robots" in early 2019,marking that coal mine robots have become an important development direction for coal mine production.As a high-risk industry,the coal industry urgently needs "machine substitution".For this reason,the positioning problem of coal mine robots must be solved first.Due to the narrow space,complex terrain,low illumination,toxic and explosive gas,high dust,high temperature,humidity,etc.,as well as the influence of multipath effects and non-visual errors,the underground coal mine mobile robot is used in the underground coal mine roadway.Accurate positioning becomes extremely difficult.In order to ensure that the robot can accurately locate in the coal mine tunnel in the harsh environment,this paper has carried out related research on the multi-sensor fusion positioning technology of the mobile robot in the coal mine.The main research work is as follows:In order to carry out related research on the multi-sensor fusion positioning technology of mobile robots in coal mines,the existing mobile robot platforms in coal mines have been designed and improved.In terms of hardware design,the control system and power drive system of the mobile robot in the coal mine have been developed,and the power drive system has been designed for explosion-proof;the wireless Mesh communication system in the coal mine has been built.In terms of software design,Rviz was used to visualize the positioning results of underground coal mine mobile robots and ultra-wideband(UWB);the software debugging and analysis of the underground wireless Mesh communication system in coal mines was carried out,and the battery power was also visualized online.In order to reduce the influence of the inaccuracy of internal and external parameters of multiple sensors on the accuracy of fusion positioning,a joint calibration of multiple sensors is required.The imaging principle and distortion model of the camera are analyzed;the reprojection error distribution interval obtained by the Kalibr calibration method is,and the average reprojection error obtained by the Matlab calibration method is 0.19 pixels.The error source of IMU is analyzed,and the IMU error model is constructed;the random error and deterministic error of IMU are respectively calibrated,and parameters such as IMU zero offset,random walk and deviation are obtained through calibration.The error source of UWB is analyzed,and the UWB calibration model is constructed and solved;the scale factor and offset of UWB are determined through calibration and linear fitting.Through the offline joint calibration of ZED/IMU/UWB,the external parameters and time drift between ZED and IMU,and the external parameters between UWB and IMU are determined.In order to solve the problems of motion blur and low illumination in the original images of coal mine tunnels collected by binocular cameras,the coal mine image processing technology was studied.In the aspect of de-motion blur,a degeneration model of motion blur images in coal mines is constructed;the blur direction and blur radius of real coal mine motion blur images are estimated,and the comparison test before and after processing multiple sets of roadway motion blur images validates the de-motion image constructed in this paper.The effectiveness of the fuzzy method.In terms of low-illuminance image processing,a global adaptive mapping function for low-illuminance images in coal mines is constructed to adaptively enhance contrast and color restoration.Through multiple sets of different roadway environmental tests,the effectiveness of the low-illuminance treatment method constructed in this paper is verified.Finally,through the comparison test of feature point extraction before and after image processing,it is verified that the coal mine motion blur and low illumination processing method constructed in this paper can effectively improve the quantity and quality of extracted feature points.In order to obtain the global positioning result of the mobile robot in the coal mine,the UWB positioning system and the pose estimation method of the mobile robot in the coal mine are studied.Analyzed the principle of UWB ranging,constructed a UWB positioning system based on EKF,and analyzed the positioning accuracy and deficiencies of the UWB positioning system for the robot through a simulated roadway test.Improved a method of UWB/IMU fusion based on graph optimization to estimate the pose of a mobile robot in coal mines.The principle of graph optimization,pose estimation method and optimization are introduced.The positioning accuracy of this method is analyzed through roadway test,and compared with The positioning results of the UWB positioning system were compared.Finally,experiments have proved that the improved UWB/IMU fusion pose estimation method based on graph optimization can realize the pose estimation of the robot,but the positioning accuracy needs to be strengthened.In order to realize the robot’s global pose estimation in underground coal mine roadways with low illumination,motion blur,low texture,repeatable texture,multipath effect,etc.,the binocular vision/IMU/UWB fusion positioning technology is studied.A Visual-IMU tightly coupled CMR-VI-SLAM positioning technology is proposed to realize local pose estimation,and its image processing thread and local pose optimization are introduced.A VI-UWB loosely coupled CMR-VIU-SLAM positioning technology is proposed to realize global pose estimation,its global factor graph model and sensor constraint factors are constructed,and the global pose graph optimization is analyzed.Through the two sets of tests in the roadway,and the comparative analysis with other positioning technologies,it is shown that compared with other vision and visual inertial positioning technologies,it can be used in underground coal mine roadways with extremely low illumination,motion blur,multipath effects,and non-line-of-sight errors In the case of unable to locate,CMR-VI-SLAM positioning technology can still be used for local pose estimation of mobile robots in coal mines;CMRVIU-SLAM positioning technology can still achieve 6-DOF global accurate pose estimation.The paper has 94 pictures,28 tables,and 151 references.