Skip to content

Latest commit

 

History

History
executable file
·
90 lines (52 loc) · 2.84 KB

README.md

File metadata and controls

executable file
·
90 lines (52 loc) · 2.84 KB
Raw

Livox-LiDAR-Camera Calibrator

In this repository, we aim to build a high-precision automatic calibration tool for Livox-LiDAR-Camera system using a printed chessboard.

Paper

For non-commercial research use. Please cite our Optics Express paper when use it, and it can be downloaded here:

  @article{RCLC,
  author = {Zhengchao Lai and Yue Wang and Shangwei Guo and Xiantong Meng and Jun Li and Wenhao Li and Shaokun Han}
  number = {10},
  pages = {16242--16263},
  publisher = {OSA},
  title = {Laser reflectance feature assisted accurate extrinsic calibration for non-repetitive scanning LiDAR and camera systems},
  volume = {30},
  month = {May},
  year = {2022},
  url = {http://opg.optica.org/oe/abstract.cfm?URI=oe-30-10-16242},
  doi = {10.1364/OE.453449}
  }

Calibration results

  • Grid fitting process:

  • Reproject results: point cloud to image ( more than 100m distance, toward pixel-wise align precision ):

  • Reproject results: image pixel map to point clouds:

Requirements

  • PCL (>1.7)
  • Eigen3(3.3.4)
  • OpenCV (>3.0)
  • ceres

Usage

  1. Configure data_root_path to be the data path in file config_real.yaml.

  2. build project:

    mkdir build && cd build

    cmake .. && make

  3. Segment the chessboard from pointcloud.

    ./BoardSegmentation

  4. Start calibrate and show the results.

    ./Calibrate

Realworld Datasets

Indoor and outdoor calibration data for MID-40 and Zed2 system can be downloded at GoogleDrive

Simulated Datasets

The complete code of simulation tool has been uploaded to Livox_Cam_Simulator. Some result as shown in the following figures.

  • The scan model of Livox LiDAR:

  • The zed camera combined with Livox LiDAR:

  • The Gazebo scene:

  • The rviz visulation:

Point clouds with reflectance intensity which mapped according to the color of the materials: