轨道车载平台可见光图像与激光点云数据

本研究展示了一套采集于雄安R1线用于列车车载环境前向感知的激光雷达点云与可见光图像数据集，数据集由森云GMSL相机与图达通Falcon激光雷达采集。传感器的频率均为10HZ，所有传感器通过GNSSPPS信号与硬件触发模块实现微秒级时间同步，空间坐标经外参标定后统一至本地东北天（ENU）坐标系。数据集覆盖了多种典型入侵场景，包括人员入侵、落石及杂物入侵等，充分体现了复杂铁路环境下入侵事件的多样性和随机性。数据集由三个完整的rosbag文件组成，每个rosbag内部同步记录了点云数据、图像数据及对应的时间戳，保证了多模态数据在时空上的精确对齐。本数据集可用于开展多源融合的路网净空安全智能感知体系与装备研究，实现对线路障碍物、周界入侵、自然灾害等威胁行车安全事件的全面感知，为保障雄安新区轨道交通的安全运营提供支撑。
在雄安R1线实验，在试验车上安装实验平台，配备两个森云GMSL相机（分辨率1920×1080）与图达通（Innovusion）Falcon激光雷达（等效150线，角分辨率0.04°，探测距离≥250m@10%反射率）。传感器的频率均为10HZ，所有传感器通过GNSSPPS信号与硬件触发模块实现微秒级时间同步，空间坐标经外参标定后统一至本地东北天（ENU）坐标系。图像与点云数据由运行于 Ubuntu20.04+ROSNoetic 的主控计算机采集，主控机器为爱视图灵CES-RUG-ORIN-701，使用 usb_cam 与 velodyne_driver 驱动节点发布话题，并通过rosbagrecord 命令将话题同步入 .bag 文件。采集试验车测试线路（雄安新区R1线）沿线的实验数据。针对雄安新区城市轨道交通规划和线路环境特点，开展多源融合的路网净空安全智能感知体系与装备研究，实现对线路障碍物、周界入侵、自然灾害等威胁行车安全事件的全面感知，为保障雄安新区轨道交通的安全运营提供支撑。数据量53.5GB。

This study presents a LiDAR point cloud and visible light image dataset collected for forward perception of the on-board train environment on Xiong'an Rail Line R1, acquired using Senyun GMSL cameras and Innovusion Falcon LiDARs. Both sensors operate at a frequency of 10 Hz, and all sensors achieve microsecond-level time synchronization via GNSS PPS signals and hardware trigger modules, with spatial coordinates unified to the local East-North-Up (ENU) coordinate system after extrinsic calibration. The dataset covers multiple typical intrusion scenarios, including personnel intrusion, falling rocks and debris intrusion, fully reflecting the diversity and randomness of intrusion events in complex railway environments. The dataset consists of three complete rosbag files, where each rosbag synchronously records point cloud data, image data and corresponding timestamps, ensuring precise spatiotemporal alignment of multimodal data. This dataset can be used to conduct research on multi-source fusion-based intelligent perception systems and equipment for road network clearance safety, enabling comprehensive perception of traffic safety threats such as line obstacles, perimeter intrusion and natural disasters, providing support for the safe operation of rail transit in Xiong'an New Area.

Field experiments were carried out on Xiong'an Rail Line R1: an experimental platform was installed on the test train, equipped with two Senyun GMSL cameras (resolution: 1920×1080) and an Innovusion Falcon LiDAR (equivalent to 150 channels, angular resolution of 0.04°, detection range ≥250 m at 10% reflectivity). Both sensors operate at 10 Hz, with microsecond-level time synchronization achieved via GNSS PPS signals and hardware trigger modules, and spatial coordinates unified to the local ENU coordinate system after extrinsic calibration. Image and point cloud data were collected by a main control computer running Ubuntu 20.04 + ROS Noetic, which is the Aiview CES-RUG-ORIN-701. The system used usb_cam and velodyne_driver driver nodes to publish topics, and synchronized the topics into .bag files via the rosbag record command. Experimental data were collected along the test route of Xiong'an Rail Line R1. Based on the planning and line environment characteristics of urban rail transit in Xiong'an New Area, research on multi-source fusion-based intelligent perception systems and equipment for road network clearance safety will be carried out, enabling comprehensive perception of traffic safety threats such as line obstacles, perimeter intrusion and natural disasters, providing support for the safe operation of rail transit in Xiong'an New Area. The total data volume of the dataset is 53.5 GB.

搜集汇总

背景与挑战

背景概述

该数据集包含雄安R1线列车车载环境前向感知的激光雷达点云与可见光图像数据，由森云GMSL相机和图达通Falcon激光雷达采集，频率为10HZ，数据量53.5GB。数据集覆盖多种典型入侵场景，适用于多源融合的路网净空安全智能感知研究。