Enhanced Calibration Accuracy for Non-Overlapping Multi-LiDAR Systems Using GPIS-based Surface Modeling

This dataset contains synchronized point cloud data from two LiDAR sensors (front and back) generated using Prescan simulation platform. It is designed for multi-LiDAR extrinsic calibration research, specifically for methods based on Gaussian Process Implicit Surface (GPIS) matching. **Dataset Contents:**- **Point Cloud Data**: Approximately 658 synchronized frames of PCD (Point Cloud Data) format from two LiDAR sensors - Front LiDAR (`lidar_front/`): ~658 PCD files, ~691 MB total - Back LiDAR (`lidar_back/`): ~658 PCD files, ~680 MB total- **Configuration Files**: - Ground segmentation parameters (`linefit_ground_removal_config.yml`) - 3D-to-2D conversion parameters (`points3d_to_laserscan_config.yml`)- **Pose Information**: Relative pose files for both LiDAR sensors in YAML format- **Documentation**: Complete usage instructions and technical analysis **Key Features:**- Synchronized multi-LiDAR point cloud sequences- Ground-truth relative poses for validation- Complete configuration files for preprocessing- Compatible with GPIS-based calibration methods- Generated from Prescan simulation for reproducible research **Use Cases:**- Multi-LiDAR extrinsic calibration algorithm development- GPIS matching method evaluation- Point cloud registration research- Sensor fusion algorithm testing- Benchmark dataset for calibration methods **Data Format:**- Point clouds: PCD format (compatible with PCL, MATLAB, ROS)- Configuration: YAML format- Poses: YAML format with 6-DOF (x, y, z, roll, pitch, yaw) **Technical Details:**- Total frames: ~658 synchronized pairs- Point cloud size: ~777 KB per file- Coordinate system: Vehicle coordinate frame- Synchronization: Time-synchronized between front and back LiDARs- File naming: 6-digit zero-padded format (000001.pcd, 000002.pcd, ...) **Compatibility:**This dataset is designed to work with the GPIS-based extrinsic calibration pipeline implemented in MATLAB. The main processing script is:```GPisMapMatching/scripts/script_estimate_extrinsic_gpisMatching_full.m``` **Processing Pipeline:**1. Ground segmentation and z/roll/pitch estimation2. 3D point cloud to 2D laser scan conversion3. GPIS matching for 2D extrinsic estimation4. Initial pose estimation using delta motion5. Final pose estimation using scan-to-map matching本数据集包含由Prescan仿真平台生成的两套同步点云数据，分别来自前向与后向激光雷达（LiDAR），专为基于高斯过程隐式曲面（Gaussian Process Implicit Surface，GPIS）匹配的多激光雷达外参标定研究设计。 **数据集内容：** - **点云数据**：来自两套激光雷达的约658帧PCD（Point Cloud Data，点云数据）格式同步文件 - 前向激光雷达（`lidar_front/`）：约658个PCD文件，总容量约691 MB - 后向激光雷达（`lidar_back/`）：约658个PCD文件，总容量约680 MB - **配置文件**： - 地面分割参数配置文件（`linefit_ground_removal_config.yml`） - 三维点云到二维激光扫描转换参数配置文件（`points3d_to_laserscan_config.yml`） - **位姿信息**：两套激光雷达的相对位姿文件，格式为YAML - **文档资料**：完整的使用说明与技术分析文档 **核心特性：** - 同步多激光雷达点云序列 - 用于标定验证的真值相对位姿 - 用于预处理的完整配置文件集 - 兼容基于GPIS的外参标定方法 - 由Prescan仿真平台生成，可支撑可复现的科研实验 **应用场景：** - 多激光雷达外参标定算法开发 - GPIS匹配类方法的性能评估 - 点云配准相关研究 - 传感器融合算法测试 - 标定方法的基准测试数据集 **数据格式：** - 点云文件：PCD格式，兼容PCL、MATLAB与ROS工具链 - 配置文件：YAML格式 - 位姿数据：YAML格式，包含6自由度参数（x、y、z、横滚、俯仰、偏航） **技术细节：** - 总帧数：约658组同步点云对 - 单文件点云大小：约777 KB - 坐标系：车辆坐标系 - 同步性：前、后激光雷达实现时间完全同步 - 文件命名规则：采用6位补零格式（如000001.pcd、000002.pcd……） **兼容性说明：** 本数据集专为适配MATLAB中实现的基于GPIS的外参标定流程设计，主处理脚本如下： GPisMapMatching/scripts/script_estimate_extrinsic_gpisMatching_full.m **数据处理流程：** 1. 地面分割与z轴、横滚、俯仰角估计 2. 三维点云到二维激光扫描的转换 3. 基于GPIS匹配的二维外参估计 4. 基于运动增量的初始位姿估计 5. 基于扫描到地图匹配的最终位姿估计