Research On Railway Obstacle Detection System Based On Laser Measurement Technology

With the construction and rapid development of high-speed railway transportation in China,railway transportation has become an indispensable basic transportation tool in China due to its long transportation distance,fast transportation speed,low transportation cost and outstanding carrying capacity.Therefore,the safety of railway transportation has attracted more and more attention.In view of the backward and low accuracy of the existing railway penetration detection methods,this paper USES the principle of Lidar ranging to carry out safety monitoring of the train's operating environment and identify the obstacles that have a serious impact on railway safety,which is of great practical significance to ensure the safe operation of railway traffic.Compared with traditional information acquisition equipment such as camera,camera,infrared,etc.,Lidar has the unique advantages of 24 hours a day and is not subject to weather interference,etc.,and it has the outstanding characteristics of high accuracy and fast speed to acquire the point cloud information of 3d target objects,which is conducive to a more accurate judgment of scene changes.Based on a detailed understanding of the development status of Lidar environment sensing technology at home and abroad,this paper makes an in-depth study on the railway obstacle detection system based on Lidar ranging principle,aiming at the functional requirements of this subject.The laser reflectance range of related objects in railway environment is analyzed and the sensor model of this system is determined.The overall layout of the system hardware is fully studied and discussed,the selection method of the system monitoring point is determined,and the overall design scheme of the railway obstacle detection system is obtained.According to the system function,the system hardware structure and interactive software are designed.Taking Lidar as the research object,the working principle,data acquisition process and data processing method of Lidar are studied.Aiming at the limitation of radar data acquisition,a method of combining Gazebo and Rviz under ROS system was proposed,and the radar data acquisition process was simulated by PCL point cloud library.The 3d modeling software is used to model the scene of railway operation,and the data acquisition of railway environment by Lidar data acquisition system is completed.On the basis of railway scene Lidar data acquisition,a series of researches on point cloud data processing are carried out.The method of obstacle detection by Lidar and the process of point cloud data processing are systematically studied.Aiming at the particularity of complex scene and uneven ground in train running environment,a ground filtering method suitable for railway special environment is put forward by studying the background difference method in 2d image processing technology.The problem of voids in the scene with dynamic objects in the background difference method in 2d computer vision is effectively avoided.After dividing the target range and filtering the collected data of radar,the detection function of Lidar scene target detection,namely railway obstacle detection,is studied.Using the difference principle of VFH eigenvalue of different objects to realize the recognition function of object types.The VFH eigenvalue database of 3d point cloud model which can be indexed by target detection system is established with kd-tree.Finally,the data processing process and the system's target detection function are tested,and the results show that the system can detect obstacles within the effective range.