qpc001 / FAST_LIO_SLAM

为FAST-LIO2中的ikd-tree部分添加中文注释

FAST_LIO_SLAM

中文注释说明

在学习FAST-LIO2的过程中，希望着重理解ikd-tree的算法哲学和代码思路，因此基于自己的理解和思考，在代码里做了相关的中文注释。注释部分难免有误，欢迎大家批评指正。

News

• Aug 2021: The Livox-lidar tests and corresponding launch files will be uploaded soon. Currenty only Ouster lidar tutorial videos had been made.

What is FAST_LIO_SLAM?

Integration of

1. FAST-LIO2 (Odometry): A computationally efficient and robust LiDAR-inertial odometry (LIO) package
2. SC-PGO (Loop detection and Pose-graph Optimization): Scan Context-based Loop detection and GTSAM-based Pose-graph optimization

Features

• An easy-to-use plug-and-play LiDAR SLAM
  • FAST-LIO2 and SC-PGO run separately (see below How to use? tab).
  • SC-PGO takes odometry and lidar point cloud topics from the FAST-LIO2 node.
  • Finally, an optimized map is made within the SC-PGO node.

How to use?

• The below commands and the launch files are made for playing the MulRan dataset, but applicable for livox lidars in the same way (you could easily make your own launch files).

    # terminal 1: run FAST-LIO2 
    mkdir -p ~/catkin_fastlio_slam/src
    cd ~/catkin_fastlio_slam/src
    git clone https://github.com/gisbi-kim/FAST_LIO_SLAM.git
    git clone https://github.com/Livox-SDK/livox_ros_driver
    cd .. 
    catkin_make
    source devel/setup.bash
    roslaunch fast_lio mapping_ouster64_mulran.launch # setting for MulRan dataset 

    # open the other terminal tab: run SC-PGO
    cd ~/catkin_fastlio_slam
    source devel/setup.bash
    roslaunch aloam_velodyne fastlio_ouster64.launch # setting for MulRan dataset 

    # open the other terminal tab
    # run file_player_mulran (for the details, refer here https://github.com/irapkaist/file_player_mulran)

Utility

• We support keyframe scan saver (as in .pcd) and provide a script reconstructs a point cloud map by merging the saved scans using the optimized poses. See here.

Example results

• Tutorial video 1 (using KAIST 03 sequence of MulRan dataset)

  • Example result captures

    

  • download the KAIST03 pcd map made by FAST-LIO-SLAM, 500MB

• Example Video 2 (Riverside 02 sequence of MulRan dataset)

  • Example result captures

    

  • download the Riverisde02 pcd map made by FAST-LIO-SLAM, 400MB

Acknowledgements

• Thanks for FAST_LIO authors.