Feature Detection and Classification in Buried Pipes using LiDAR: Dataset with Camera-LiDAR Synchronisation

LiDAR–Camera SynchronisationThis repository contains data collected from a Minicam crawler robot navigating confined pipe environments, equipped with a front-facing camera and a 2D LiDAR sensor. The pipe network includes two 2.5-meter concrete pipe sections, two 2-meter clay pipe sections and two 1-meter manholes. The pipe network has a total length of 11 meters and a diameter of 300 mm.The collection was created to address challenges of accurate navigation, mapping, feature detection and classification within feature-sparse pipe networks.The dataset includes:Visual data captured during robot navigation (MP4 videos).2D LiDAR data provided both as raw ROS2 bag recordings and pre-extracted CSV scans.Ground-truth annotations describing pipe structural features (joints, manholes, clean path).Algorithm predictions generated by statistical models presented in our CCWI paper cited below.Synchronised outputs combining camera frames with corresponding LiDAR scans.A README file describing the repository structure, dataset contents, and usage instructions.This repository also provides tools and scripts for synchronising 2D LiDAR scans with camera video for underground crawler inspections. The goal is to align both camera and LiDAR streams in time.CitationIf you use this dataset, please cite the CCWI 2025 paper:Karnezis, Aristeidis; Worley, Rob; Blight, Andy; Anderson, Sean; Horoshenkov, Kirill; Mihaylova, Lyudmila (2025). Feature Detection and Classification in Buried Pipes using LiDAR Technology. The University of Sheffield. Conference contribution. https://doi.org/10.15131/shef.data.29920931激光雷达（LiDAR）-相机同步 本仓库收录了在受限管道环境中巡检的迷你相机型管道爬行机器人采集的数据，该机器人搭载前向相机与二维激光雷达（LiDAR）传感器。本次搭建的管道网络包含两段2.5米长的混凝土管段、两段2米长的黏土管段以及两处1米长的检查井，管道网络总长度为11米，管径为300毫米。 本数据集的采集旨在解决特征稀疏的管道网络内精确导航、建图、特征检测与分类的技术难题。 数据集包含以下内容： 机器人巡检过程中采集的视觉数据（MP4视频文件） 二维激光雷达（LiDAR）数据，同时提供ROS2包格式的原始记录与预提取的CSV格式扫描文件 描述管道结构特征（接缝、检查井、通畅路径）的真值标注 由我们在下文引用的CCWI 2025会议论文中提出的统计模型生成的算法预测结果 将相机帧与对应激光雷达扫描数据同步整合的输出文件 一份README文件，用于说明仓库结构、数据集内容与使用方法 本仓库还提供适用于地下管道爬巡检测场景的工具与脚本，用于实现二维激光雷达扫描数据与相机视频的时间同步，目标是将相机与激光雷达数据流进行时间对齐。 引用说明 若您使用本数据集，请引用2025年CCWI会议论文： Karnezis, Aristeidis; Worley, Rob; Blight, Andy; Anderson, Sean; Horoshenkov, Kirill; Mihaylova, Lyudmila (2025). 基于激光雷达（LiDAR）技术的地下管道特征检测与分类. 谢菲尔德大学. 会议论文. https://doi.org/10.15131/shef.data.29920931