vslam

Published: by Creative Commons Licence

Visual Simultaneous Localization and Mapping,

Visual SLAM开源框架分类

Created 2019.01.01 by William Yu; Last modified: 2019.01.01-V1.0.0

Contact: windmillyucong@163.com

Copyright ©2018 William Yu. All rights reserved.


分类方法

按照视觉传感器分类:

  1. 单目摄像头(Monocular Camera)
  2. 双目摄像头(Binocular Camera)
  3. 深度摄像头(RGB-D Camera)

根据前端分类:

  1. 非直接法(Indirect)(一般为基于特征(Feature-Based),比如特征点,线面特征)
  2. 直接法(Direct)

根据构建地图的稀疏程度分类:

  1. 稀疏(Sparse)
  2. 稠密(Dense)

  • Sparse + Indirect:

    monoSLAM

    A. Davison, I. Reid, N. Molton, and O. Stasse. MonoSLAM: Real-time single camera SLAM. Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 29, 2007. 2

    PTAM

    G. Klein and D. Murray. Parallel tracking and mapping for small AR workspaces. In International Symposium on Mixed and Augmented Reality (ISMAR), 2007. 2

    ORB-SLAM

    R. Mur-Artal, J. Montiel, and J. Tardos. ORB-SLAM: a versatile and accurate monocular SLAM system. Transactions on Robotics, 31(5):1147–1163, 2015. 2, 11

  • Dense + Indirect

    L. Valgaerts, A. Bruhn, M. Mainberger, and J. Weickert. Dense versus sparse approaches for estimating the fundamental matrix. International Journal of Computer Vision (IJCV), 96(2):212–234, 2012. 2

    R. Ranftl, V. Vineet, Q. Chen, and V. Koltun. Dense monocular depth estimation in complex dynamic scenes. In International Conference on Computer Vision and Pattern Recognition (CVPR), 2016. 2

  • Dense + Direct:

    DTAM

    R. Newcombe, S. Lovegrove, and A. Davison. DTAM: Dense tracking and mapping in real-time. In International Conference on Computer Vision (ICCV), 2011. 2, 3

    J. St¨uhmer, S. Gumhold, and D. Cremers. Real-time dense geometry from a handheld camera. In Pattern Recognition (DAGM), 2010. 2, 3

    LSD-SLAM

    J. Engel, T. Sch¨ops, and D. Cremers. LSD-SLAM: Largescale direct monocular SLAM. In European Conference on Computer Vision (ECCV), 2014. 2, 11

  • Sparse + Direct

    DSO

    J. Engel, V. Usenko and D. Cremers, A Photometrically Calibrated Benchmark For Monocular Visual Odometry, arXiv:1607.02555, July 2016.

    J. Engel, V. Koltun and D. Cremers, Direct Sparse Odometry, arXiv:1607.02565, July 2016.

VSLAM Sensor Time 优点 缺点 Front-end Back-end 特殊点
MonoSLAM12 单目 2007   应用场景非常窄,已经停止开发 稀疏点 EKF 第一个可实时运行
PTAM3 单目 2007     关键帧 非线性优化 1.实现了跟踪与建图过程的并行化
2.第一个使用非线性优化方案
3.引入关键帧机制
ORB-SLAM4 单目为主 2015 1.支持单目双目RGBD多种相机,是一套完善的开源算法
2.计算ORB特征点,包括视觉里程计与回环检测ORB字典。ORB特征不像SURF或SIFT费时,可在CPU上实时计算;相比Harris等简单角点特征,又具有良好的旋转和缩放不变性;ORB提供描述子,使在大范围运动时能够回环检测和重定位。
3. ORB的回环检测是它的亮点,优秀的回环检测算法保证了ORB-SLAM有效的防止累计误差,并且在丢失后还能迅速找回。为此,ORB-SLAM在运行前需要加载一个很大的ORB字典文件
由于整个 SLAM系统采用特征点进行计算,对每幅图像都计算一遍ORB特征,是非常耗时的;ORB-SLAM的三线程结构给CPU带来了较大的负担;ORB-SLAM的建图为稀疏矩阵点,目前还没有开放存储和读取地图后重新定位的功能,稀疏特征点地图只能满足我们对定位的需求,而无法提供导航、避障、交互等功能。      
ORB-SLAM2 单目            
LSD-SLAM 单目为主 2014 半稠密 对相机曝光异常敏感,快速运动容易跟丢   图优化  
SVO 单目 2014 速度快 舍弃了后端优化和回环检测环节,没有基本的建图功能 稀疏直接法 深度滤波器  
DTAM RGBD            
DVO RGBD            
DSO              
RTAB-MAP 双目+RGBD   支持RGB-D和双目集成度比较高的相机        
OKVIS 多目+IMU       关键帧 非线性优化  
Elastic Fusion RGBD     构建大场景地图时不适用,代码优化较差      
ROVIO 单目+IMU   计算量小 没有闭环检测和地图构建部分   EKF  
RGBD-SLAM              

License

CC0

  1. Davison, A. J., et al. "MonoSLAM: real-time single camera SLAM. " IEEE Transactions on Pattern Analysis & Machine Intelligence29.6(2007):1052. 

  2. Davison, Andrew J. "Real-Time Simultaneous Localisation and Mapping with a Single Camera." IEEE International Conference on Computer Vision IEEE Computer Society, 2003:1403. 

  3. Murray, Dw, and G. Klein. "Parallel tracking and mapping for small AR workspaces." (2007):1-10. 

  4. Mur-Artal, Raúl, J. M. M. Montiel, and J. D. Tardós. "ORB-SLAM: A Versatile and Accurate Monocular SLAM System." IEEE Transactions on Robotics 31.5(2015):1147-1163.