{"id":28418882,"url":"https://github.com/engcang/fast-lio-sam-qn","last_synced_at":"2025-06-25T21:31:04.355Z","repository":{"id":181684759,"uuid":"665212843","full_name":"engcang/FAST-LIO-SAM-QN","owner":"engcang","description":"A SLAM implementation combining FAST-LIO2 with pose graph optimization and loop closing based on Quatro and Nano-GICP","archived":false,"fork":false,"pushed_at":"2024-10-27T13:56:25.000Z","size":956,"stargazers_count":281,"open_issues_count":2,"forks_count":21,"subscribers_count":6,"default_branch":"master","last_synced_at":"2025-06-04T19:30:43.975Z","etag":null,"topics":["lidar-inertial-odometry","lidar-slam","loop-closure","odometry","pose-estimation","slam"],"latest_commit_sha":null,"homepage":"","language":"C++","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/engcang.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2023-07-11T17:34:01.000Z","updated_at":"2025-05-28T09:07:43.000Z","dependencies_parsed_at":"2023-12-20T13:18:49.388Z","dependency_job_id":"927c2216-5338-4fd3-807d-8952e2dd9873","html_url":"https://github.com/engcang/FAST-LIO-SAM-QN","commit_stats":null,"previous_names":["engcang/fast-lio-sam-qn"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/engcang/FAST-LIO-SAM-QN","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/engcang%2FFAST-LIO-SAM-QN","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/engcang%2FFAST-LIO-SAM-QN/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/engcang%2FFAST-LIO-SAM-QN/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/engcang%2FFAST-LIO-SAM-QN/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/engcang","download_url":"https://codeload.github.com/engcang/FAST-LIO-SAM-QN/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/engcang%2FFAST-LIO-SAM-QN/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":261955853,"owners_count":23235963,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["lidar-inertial-odometry","lidar-slam","loop-closure","odometry","pose-estimation","slam"],"created_at":"2025-06-04T13:20:43.660Z","updated_at":"2025-06-25T21:31:04.343Z","avatar_url":"https://github.com/engcang.png","language":"C++","funding_links":[],"categories":[],"sub_categories":[],"readme":"# FAST-LIO-SAM-QN\n+ This repository is a SLAM implementation combining [FAST-LIO2](https://github.com/hku-mars/FAST_LIO) with pose graph optimization and loop closing based on [Quatro](https://quatro-plusplus.github.io/) and [Nano-GICP module](https://github.com/engcang/nano_gicp)\n    + [Quatro](https://quatro-plusplus.github.io/) - fast, accurate and robust global registration which provides great initial guess of transform\n    + [Quatro module](https://github.com/engcang/quatro) - `Quatro` as a module, can be easily used in other packages\n    + [Nano-GICP module](https://github.com/engcang/nano_gicp) - fast ICP combining [FastGICP](https://github.com/SMRT-AIST/fast_gicp) + [NanoFLANN](https://github.com/jlblancoc/nanoflann)\n+ Note: similar repositories already exist\n    + [FAST_LIO_LC](https://github.com/yanliang-wang/FAST_LIO_LC): FAST-LIO2 + SC-A-LOAM based SLAM\n    + [FAST_LIO_SLAM](https://github.com/gisbi-kim/FAST_LIO_SLAM): FAST-LIO2 + ScanContext based SLAM\n    + [FAST_LIO_SAM](https://github.com/kahowang/FAST_LIO_SAM): FAST-LIO2 + LIO-SAM (not modularized)\n    + [FAST_LIO_SAM](https://github.com/engcang/FAST-LIO-SAM): FAST-LIO2 + LIO-SAM (modularized)\n+ Note2: main code (PGO) is modularized and hence can be combined with any other LIO / LO\n    + This repo is to learn GTSAM myself!\n    + and as GTSAM tutorial for beginners - [GTSAM 튜토리얼 한글 포스팅](https://engcang.github.io/2023/07/15/gtsam_tutorial.html)\n## Video clip - https://youtu.be/MQ8XxRY472Y\n\u003cbr\u003e\n\n## Computational complexity \u003cbr\u003ein KITTI seq. 05 with i9-10900k CPU\n+ FAST-LIO-SAM: max 118% CPU usage, 125 times of ICP, 124.9ms consumption on average\n+ FAST-LIO-SAM-N (only Nano-GICP): max 164% CPU usage, 130 times of ICP, 61.9ms consumption on average\n+ FAST-LIO-SAM-QN:\n    + Advanced matching - max 325% CPU usage, 85 times of ICP, 140ms consumption on average\n    + Optimized matching (with max 200 correspondences downsampling) - max 569% CPU usage, 90 times of ICP, 128.6ms consumption on average\n+ Note: `loop_timer_func` runs at fixed `basic/loop_update_hz`. So how many times of ICP occurred can be different depending on the speed of matching methods.\n\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"imgs/fast1.png\" height=\"250\"/\u003e\n  \u003cimg src=\"imgs/sam1.png\" height=\"250\"/\u003e\n  \u003cimg src=\"imgs/qn.png\" height=\"250\"/\u003e\n  \u003cbr\u003e\n  \u003cem\u003eKITTI seq 05 top view - (left): FAST-LIO2 (middle): FAST-LIO-SAM (bottom): FAST-LIO-SAM-QN\u003c/em\u003e\n\u003c/p\u003e\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"imgs/fast2.png\" width=\"500\"/\u003e\n  \u003cimg src=\"imgs/sam2.png\" width=\"500\"/\u003e\n  \u003cimg src=\"imgs/qn_side.png\" width=\"500\"/\u003e\n  \u003cbr\u003e\n  \u003cem\u003eKITTI seq 05 side view - (top): FAST-LIO2 (middle): FAST-LIO-SAM (bottom): FAST-LIO-SAM-QN\u003c/em\u003e\n\u003c/p\u003e\n\n\u003cbr\u003e\n\n## Dependencies\n+ `C++` \u003e= 17, `OpenMP` \u003e= 4.5, `CMake` \u003e= 3.10.0, `Eigen` \u003e= 3.2, `Boost` \u003e= 1.54\n+ `ROS`\n+ [`GTSAM`](https://github.com/borglab/gtsam)\n    ```shell\n    wget -O gtsam.zip https://github.com/borglab/gtsam/archive/refs/tags/4.1.1.zip\n    unzip gtsam.zip\n    cd gtsam-4.1.1/\n    mkdir build \u0026\u0026 cd build\n    cmake -DGTSAM_BUILD_WITH_MARCH_NATIVE=OFF -DGTSAM_USE_SYSTEM_EIGEN=ON ..\n    sudo make install -j16\n    ```\n+ [`Teaser++`](https://github.com/MIT-SPARK/TEASER-plusplus)\n    ```shell\n    git clone https://github.com/MIT-SPARK/TEASER-plusplus.git\n    cd TEASER-plusplus \u0026\u0026 mkdir build \u0026\u0026 cd build\n    cmake .. -DENABLE_DIAGNOSTIC_PRINT=OFF\n    sudo make install -j16\n    sudo ldconfig\n    ```\n+ `tbb` (is used for faster `Quatro`)\n    ```shell\n    sudo apt install libtbb-dev\n    ```\n\n## How to build\n+ Get the code and then build the main code.\n    ```shell\n    cd ~/your_workspace/src\n    git clone https://github.com/engcang/FAST-LIO-SAM-QN --recursive\n\n    cd ~/your_workspace\n    # nano_gicp, quatro first\n    catkin build nano_gicp -DCMAKE_BUILD_TYPE=Release\n    # Note the option!\n    catkin build quatro -DCMAKE_BUILD_TYPE=Release -DQUATRO_TBB=ON -DQUATRO_DEBUG=OFF\n    catkin build -DCMAKE_BUILD_TYPE=Release\n    . devel/setup.bash\n    ```\n\n## How to run\n+ Then run (change config files in third_party/`FAST_LIO`)\n    ```shell\n    roslaunch fast_lio_sam_qn run.launch lidar:=ouster\n    roslaunch fast_lio_sam_qn run.launch lidar:=velodyne\n    roslaunch fast_lio_sam_qn run.launch lidar:=livox\n    ```\n* In particular, we provide a preset launch option for specific datasets:\n    ```shell\n    roslaunch fast_lio_sam_qn run.launch lidar:=kitti\n    roslaunch fast_lio_sam_qn run.launch lidar:=mulran\n    roslaunch fast_lio_sam_qn run.launch lidar:=newer-college20\n    ```\n\n## Structure\n+ odomPcdCallback\n    + pub realtime pose in corrected frame\n    + keyframe detection -\u003e if keyframe, add to pose graph + save to keyframe queue\n    + pose graph optimization with iSAM2\n+ loopTimerFunc\n    + process a saved keyframe\n        + detect loop -\u003e if loop, add to pose graph\n+ visTimerFunc\n    + visualize all **(Note: global map is only visualized once uncheck/check the mapped_pcd in rviz to save comp.)**\n\n\u003cbr\u003e\n\n## Memo\n+ `Quatro` module fixed for empty matches\n+ `Quatro` module is updated with `optimizedMatching` which limits the number of correspondences and increased the speed\n\n\u003cbr\u003e\n\n## License\n\u003ca rel=\"license\" href=\"http://creativecommons.org/licenses/by-nc-sa/4.0/\"\u003e\u003cimg alt=\"Creative Commons License\" style=\"border-width:0\" src=\"https://i.creativecommons.org/l/by-nc-sa/4.0/88x31.png\" /\u003e\u003c/a\u003e\n\nThis work is licensed under a [Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License](http://creativecommons.org/licenses/by-nc-sa/4.0/)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fengcang%2Ffast-lio-sam-qn","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fengcang%2Ffast-lio-sam-qn","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fengcang%2Ffast-lio-sam-qn/lists"}