README.md

metadata

license: mit

Mandeye Mapping – Loop Closure dataset

Handheld LiDAR + IMU + camera dataset recorded with a Mandeye scanner in an urban courtyard, with a loop in the trajectory. It comes with LiDAR-inertial odometry results from HDMapping (v0.104.0) and a LiDAR–camera calibration, so you can colorize the point cloud from the camera images.

▶️ Video: https://youtu.be/jxjWX8dq30Y

Contents

Path	Description
lidar*.laz, lidar*.sn	Raw LiDAR chunks (point clouds + serial numbers)
imu*.csv	IMU data (gyro, accelerometer, timestamps)
status*.json	Scanner status snapshots
CAMERA_0_blurred/	Camera images (cam0_<timestamp>.jpg), faces/plates blurred
extrinsics.json	Camera intrinsics (fx, fy, cx, cy, distortion) and T_lidar_to_camera_4x4 extrinsics
lio_result_0/	HDMapping LiDAR-inertial odometry output: scan_lio_*.laz, trajectory_lio_*.csv, session.mjs, params .toml

Usage

1. Download the dataset:

   git lfs install
   git clone https://huggingface.co/datasets/MichalPelka/MandeyeMappingLoopClosure
   

2. Get HDMapping (releases).
3. (Optional) Re-run LiDAR odometry on the raw data with HDMapping's lidar_odometry_step_1, or use the provided lio_result_0/.
4. Open the Trajectory Viewer and in the side panel:
   • Session → LIO result directory: ./lio_result_0, leave CAMERA_0 directory empty (auto-detects CAMERA_0_blurred) → Load session → Load cloud
   • Calibration → Calibration JSON: extrinsics.json → Load calibration
   • Export → Export colored LAZ / Export colored E57

Links

• Video: https://youtu.be/jxjWX8dq30Y
• HDMapping: https://github.com/MapsHD/HDMapping
• Mandeye controller: https://github.com/JanuszBedkowski/mandeye_controller