资料 | 2021年开源SLAM算法集锦-技术圈

本文汇总2021年公布的SLAM相关工作，后台回复“2021SLAM”可获得本文全部论文。

1.TANDEM：Tracking and Dense Mapping in Real-time using Deep Multi-view Stereo

慕尼黑工业大学Daniel Cremers团队，实时单目跟踪稠密建图纯视觉SLAM，采用Realsense D455(深度传感器+IMU，但只用RGB)
项目地址：https://vision.in.tum.de/research/vslam/tandem
论文地址：https://arxiv.org/pdf/2111.07418.pdf
源码地址：https://github.com/tum-vision/tandem

2.MonoRec ：Semi-Supervised Dense Reconstruction in Dynamic Environments from a Single Moving Camera

慕尼黑工业大学Daniel Cremers团队，半监督单目稠密重建纯视觉SLAM
项目地址：https://vision.in.tum.de/research/monorec
论文地址：https://arxiv.org/pdf/2011.11814.pdf
源码地址：https://github.com/Brummi/MonoRec

3.Range-MCL:Range Image-based LiDAR Localization for Autonomous Vehicles

波恩大学 Cyrill Stachniss团队，3D LiDAR户外激光SLAM，采用Passion表面重建和蒙特卡洛定位框架
项目地址：https://www.ipb.uni-bonn.de/research/，https://www.ipb.uni-bonn.de/data-software/
论文地址：https://arxiv.org/pdf/2105.12121.pdf
源码地址：https://github.com/PRBonn/range-mcl

4.MULLS:Versatile LiDAR SLAM via Multi-metric Linear Least Square

ETH苏黎世联邦理工学院、EPFL洛桑联邦理工学院、禾赛科技，激光SLAM
项目地址：https://baug.ethz.ch/en/，https://www.hesaitech.com/zh/，https://sti.epfl.ch
论文地址：https://arxiv.org/pdf/2102.03771.pdf
源码地址：https://github.com/YuePanEdward/MULLS

5.LiLi-OM:Towards High-Performance Solid-State-LiDAR-Inertial Odometry and Mapping

KIT德国卡尔斯鲁厄理工学院，实时紧耦合激光雷达惯性里程计SLAM，特征提取参考固态激光雷达 Livox Horizon 与机械激光雷达Velodyne A-LOAM( HKUST-Aerial-Robotics )，可先参考开源VINS-Fusion(https://github.com/HKUST-Aerial-Robotics/VINS-Fusion)和 LIO-mapping(https://github.com/hyye/lio-mapping)。
项目地址：https://isas.iar.kit.edu/
论文地址：https://arxiv.org/pdf/2010.13150v3
源码地址：https://github.com/KIT-ISAS/lili-om

6.FAST-LIO2：Fast Direct LiDAR-inertial Odometry

FAST-LIO: A Fast, Robust LiDAR-inertial Odometry Package by Tightly-Coupled Iterated Kalman Filter
香港大学张富团队，在FAST-LIO(高效鲁棒性LiDAR、惯性里程库，融合LiDAR特征点和IIMU数据，紧耦合快速EKF迭代)基础上采用ikd-Tree(https://github.com/hku-mars/ikd-Tree)增量建图，原始LiDAR点直接计算里程，支持外部IMU，并支持ARM平台。
项目地址：https://mars.hku.hk
论文地址：https://arxiv.org/pdf/2107.06829v1.pdf，

https://arxiv.org/pdf/2010.08196v3.pdf
源码地址：https://github.com/hku-mars/FAST_LIO
相关工作：
ikd-Tree: A state-of-art dynamic KD-Tree for 3D kNN search. https://github.com/hku-mars/ikd-Tree
IKFOM: A Toolbox for fast and high-precision on-manifold Kalman filter. https://github.com/hku-mars/IKFoM
UAV Avoiding Dynamic Obstacles: One of the implementation of FAST-LIO in robot's planning.https://github.com/hku-mars/dyn_small_obs_avoidance
R2LIVE: A high-precision LiDAR-inertial-Vision fusion work using FAST-LIO as LiDAR-inertial front-end.https://github.com/hku-mars/r2live
UGV Demo: Model Predictive Control for Trajectory Tracking on Differentiable Manifolds.https://www.youtube.com/watch?v=wikgrQbE6Cs
FAST-LIO-SLAM: The integration of FAST-LIO with Scan-Context loop closure module.https://github.com/gisbi-kim/FAST_LIO_SLAM
FAST-LIO-LOCALIZATION: The integration of FAST-LIO with Re-localization function module.https://github.com/HViktorTsoi/FAST_LIO_LOCALIZATION

7.R3LIVE:A Robust, Real-time, RGB-colored, LiDAR-Inertial-Visual tightly-coupled state Estimation and mapping package

香港大学张富团队，在R2LIVE(FAST-LIO与VIO)基础上，LiDAR、惯导、视觉多传感器融合SLAM
项目地址：https://mars.hku.hk/
论文地址：https://arxiv.org/pdf/2109.07982.pdf
源码地址：https://github.com/hku-mars/r3live
相关工作：

数据集：https://github.com/ziv-lin/r3live_dataset

R2LIVE: A robust, real-time tightly-coupled multi-sensor fusion package.https://github.com/hku-mars/r2live
FAST-LIO: A computationally efficient and robust LiDAR-inertial odometry package.https://github.com/hku-mars/FAST_LIO
ikd-Tree: A state-of-art dynamic KD-Tree for 3D kNN search.https://github.com/hku-mars/ikd-Tree LOAM-Livox: A robust LiDAR Odometry and Mapping (LOAM) package for Livox-LiDAR.https://github.com/hku-mars/loam_livox
openMVS: A library for computer-vision scientists and especially targeted to the Multi-View Stereo reconstruction community.https://github.com/cdcseacave/openMVS
VCGlib: An open source, portable, header-only Visualization and Computer Graphics Library.https://github.com/cnr-isti-vclab/vcglib
CGAL: A C++ Computational Geometry Algorithms Library.https://www.cgal.org/,https://github.com/CGAL/cgal

8.GVINS: tightly coupled GNSS-visual-inertial fusion for smooth and consistent state estimation

香港科技大学沈邵劼团队，之前开源VINS-Mono(https://github.com/HKUST-Aerial-Robotics/VINS-Mono)，VINS-Fusion(https://github.com/HKUST-Aerial-Robotics/VINS-Fusion)，GVINS是基于GNSS、视觉、惯导紧耦合多传感器融合平滑一致状态估计。
项目地址：https://uav.hkust.edu.hk
论文地址：https://arxiv.org/pdf/2103.07899.pdf
源码地址：https://github.com/HKUST-Aerial-Robotics/GVINS
相关资源：http://www.rtklib.com/ 系统框架及VIO部分采用VINS-Mono，相机建模采用camodocal(https://github.com/hengli/camodocal)，ceres(http://ceres-solver.org/)优化。
RTKLIB: An Open Source Program Package for GNSS Positioning，An Open Source Program Package for GNSS Positioning

9.LVI-SAM:Tightly-coupled Lidar-Visual-Inertial Odometry via Smoothing and Mapping

MIT麻省理工学院TixiaoShan(之前开源LIO-SAM，https://github.com/TixiaoShan/LIO-SAM)，激光、视觉、惯性紧耦合多传感器融合SLAM，里程计建图系统层联合LIO-SAM与Vins-Mono优势，依赖与ROS、gtsam、Ceres库。
项目地址：https://git.io/lvi-sam，https://dusp.mit.edu/，https://senseable.mit.edu/，https://www.ams-institute.org/
论文地址：https://arxiv.org/pdf/2104.10831.pdf
源码地址：https://github.com/TixiaoShan/LVI-SAM

10.DSP-SLAM:Object Oriented SLAM with Deep Shape Priors

伦敦大学，基于ORB-SLAM2，面向对象语义SLAM
项目地址：https://jingwenwang95.github.io/dsp-slam
论文地址：https://arxiv.org/pdf/2108.09481v2.pdf
源码地址：https://github.com/JingwenWang95/DSP-SLAM

11. UV-SLAM: Unconstrained Line-based SLAM Using Vanishing Points for Structural Mapping

KAIST韩国科学技术院，采用消隐点实现无约束线特征结构化建图，克服传统线重投影测量模型中仅利用 Plücker坐标线法向量。
论文地址：https://arxiv.org/pdf/2112.13515.pdf
源码地址：https://github.com/url-kaist/UV-SLAM，源码即将上传
相关研究：Avoiding Degeneracy for Monocular Visual SLAM with Point and Line Features
ALVIO: Adaptive Line and Point Feature-Based Visual Inertial Odometry for Robust Localization in Indoor Environments，源码未上传https://github.com/ankh88324/ALVIO

12. Autonomous Navigation System from Simultaneous Localization and Mapping

克拉克森大学，基于slam室内导航软件架构，应用于智能轮椅
项目地址：无
论文地址：https://arxiv.org/pdf/2112.07723.pdf
源码地址：https://github.com/michealcarac/VSLAM-Mapping

13.MegBA: A High-Performance and Distributed Library for Large-Scale Bundle Adjustment

爱丁堡大学、旷视科技，大规模BA算法，GPU分布式计算
项目地址：无
论文地址：https://arxiv.org/pdf/2112.01349v2.pdf
源码地址：https://github.com/MegviiRobot/MegBA

14. Fast Direct Stereo Visual SLAM

明尼苏达大学，快速准确立体视觉SLAM，不依赖于特征探测与匹配。作者从单目DSO扩展到双目系统，通过3D点最小光度误差优化双目配置尺度。
项目地址：无
论文地址：https://arxiv.org/pdf/2112.01890.pdf
源码地址：https://github.com/IRVLab/direct_stereo_slam
相关工作：Direct Sparse Odometry，A Photometrically Calibrated Benchmark For Monocular Visual Odometry，https://github.com/JakobEngel/dso