Detection Robot Localization With Odometry And Lidar Fusion In Tunnel

In view of the current safety hazards existing in the tailings reservoir flood drainage tunnel,a detection robot was developed to detect safety hazards,and the position of hazards was determined by the precise localization of the robot.In order to accurately determine the pose of the robot in the tunnel,designing a localization method,firstly using improved UMBmark to calibrate systematic parameters of the odometry,then using the extended Kalman filter(EKF)based SLAM(Simultaneous Localization and Mapping)to fuse the odometry and lidar.Firstly,the mathematical model of the robot localization system is established,including the robot and tunnel environment coordinate system,the sensor model,the tunnel environment landmark model and the velocity kinematics model of the robot.Simulink is used to simulate the velocity kinematics model.The odometry arc kinematics model can obtain more accurate robot Localization.The odometry systematic parameters are calibrated to reduce the robot systematic error.The bidirectional square path calibration size of the detection robot is designed and the UMBmark calibration algorithm is improved.It is verified by experiments that the improved UMBmark calibration algorithm can reduce the systematic error of the odometry more.Secondly,according to the designed EKF-SLAM fusion localization algorithm,four key implementation steps are expounded,and the EKF-SLAM fusion localization of the robot in the local environment of the tunnel is simulated by Matlab,and the feasibility of the algorithm is verified.Finally,The influence of localization uncertainty on localization error is analyzed.The detection robot platform and localization system were built.In the actual tunnel environment,a 100-meter linear path localization experiment based on odometry and a 50 m×2 m gate-shaped path localization experiment were carried out.The experimental results show that after the calibrated path size was designed and the systematic parameters of the odometry were calibrated by the improved UMBmark algorithm.the robot's 100-meter linear localization error is 4.32%.Compared to before the calibration,the odometry localization accuracy is improved by 49.88%,and the robot gate-shaped path odometry localization accuracy is improved by 55.91%.In ROS(Robot Operating System),the environment with local cavity in the tunnel is simulated,and the EKF-SLAM fusion localization simulation experiment which is closer to the real situation is carried out,which proves the feasibility of the fusion localization algorithm in the actual environment.Finally,the EKF-SLAM fusion localization experiment of the detection robot was carried out in the simulated tunnel environment,and 20 fusion localization experiments were carried out in the environment.The repeated localization accuracy of the robot was tested,and the average distance error of the detection robot was less than 15.27 cm,the average heading error is less than 2.67?,within which the tunnel detection task can be completed.