Research on Initialization Algorithm for Visual-Inertial SLAM System

doi:10.3778/j.issn.1673-9418.2005043

Abstract

Abstract:

Monocular vision and inertial simultaneous localization and mapping (SLAM) system is becoming more and more popular in practical engineering applications because it can achieve the complementarity in use scenarios and lower hardware cost. Recent results have shown that optimization-based fusion strategies outperform filtering strategies. However, the optimization-based SLAM algorithm of vision inertial navigation fusion is highly nonlinear, and its performance highly depends on the accuracy of the estimation of the initial parameters of the system state. The inertial measurement unit needs acceleration excitation, which means that it cannot start from the static state, but must start from the unknown motion state. Therefore, accurate estimation of the initial state is the key to the high robustness of the algorithm and the first step of the vision inertial fusion algorithm. By analyzing the pre integration algorithm of inertial measurement unit, an initialization estimation system based on convex optimization is derived, and the initial states are solved jointly considering the constraints of the gravity acceleration. More importantly, a novel method is proposed to determine the termination condition of the initialization algorithm by measuring the estimation effect with Fisher information, which improves the accuracy of the algorithm and shortens the initialization time. Experiments on Euroc dataset show that the new algorithm has a more precise and robust initial state.

Key words: simultaneous localization and mapping (SLAM), visual inertial alignment, preintegration, inertial navigation, Fisher information

摘要：

单目视觉系统融合惯性测量单元的同时定位与地图构建（SLAM）系统，能实现在使用场景上的互补，以及较低的硬件成本，在实际工程应用中越来越受到青睐。最近的研究表明，基于优化的SLAM算法性能优于基于滤波的SLAM算法。基于优化的视觉惯导融合SLAM算法具有高度非线性化的特点，其性能高度依赖于系统初始状态估计的准确性；惯性测量单元需要加速度激励，这意味着不能从静止状态启动，而必须从未知的运动状态启动，因此如何确定这一未知的初始状态显得尤为重要。综上可知，对初始状态准确的估计是SLAM算法具有高精确鲁棒性的关键，也是视觉惯性融合算法的第一步。通过对惯性测量单元预积分算法的分析，推导出一种凸优化的初始化估计系统。在综合考虑了重力加速度的约束条件下，对各初始状态进行联合求解。更重要的是，提出了一种新颖的方法，即通过费歇尔信息衡量估计效果的好坏来确定初始化算法的终止条件，提高算法精确度的同时也缩短了初始化的时间。在Euroc数据集上的实验表明，该算法具有更高精确鲁棒的初始状态。

关键词: 同时定位与地图构建（SLAM）, 视觉惯性对准, 预积分, 惯性导航, 费歇尔信息

LIU Gang, GE Hongwei. Research on Initialization Algorithm for Visual-Inertial SLAM System[J]. Journal of Frontiers of Computer Science and Technology, 2021, 15(8): 1546-1554.

刘刚, 葛洪伟. 视觉惯导SLAM初始化算法研究[J]. 计算机科学与探索, 2021, 15(8): 1546-1554.

References

[1] KLEIN G, MURRAY D W. Parallel tracking and mapping for small AR workspaces[C]//Proceedings of the 6th IEEE/ACM International Symposium on Mixed and Augmented Reality, Nara, Nov 13-16, 2007. Washington: IEEE Computer Society, 2007: 225-234.
[2] MOURIKIS A I, ROUMELIOTIS S I. A multi-state cons-traint Kalman filter for vision-aided inertial navigation[C]// Proceedings of the 2007 IEEE International Conference on Robotics and Automation, Roma, Apr 10-14, 2007. Pisca-taway: IEEE, 2007: 3565-3572.
[3] MUR-ARTAL R, MONTIEL J M M, TARDOS J D. ORB-SLAM: a versatile and accurate monocular SLAM system[J]. IEEE Transactions on Robotics, 2015, 31(5): 1147-1163.
[4] QIN T, SHEN S J. Robust initialization of monocular visual-inertial estimation on aerial robots[C]//Proceedings of the 2017 IEEE/RSJ International Conference on Intelligent Ro-bots & Systems, Vancouver, Sep 24-28, 2017. Piscataway: IEEE, 2017: 4225-4232.
[5] LEUTENEGGER S, FURGALE P T, RABAUD V, et al. Keyframe-based visual-inertial SLAM using nonlinear opti-mization[C]//Proceedings of the Robotics: Science and Systems IX, Berlin, Jun 24-28, 2013: 1-8.
[6] JONES E S, SOATTO S. Visual-inertial navigation, mapping and localization: a scalable real-time causal approach[J]. The International Journal of Robotics Research, 2011, 30(4): 407-430.
[7] LEUTENEGGER S, LYNEN S, FURGALE P, et al. Keyframe-based visual-inertial odometry using nonlinear optimization[J]. International Journal of Robotics Research, 2015, 34:314-334.
[8] GEORGE M, TARDIF J P, KELLY A. Visual and inertial odometry for a disaster recovery humanoid[M]//Field and Service Robotics. Berlin, Heidelberg: Springer, 2015: 501-514.
[9] WU K J, AHMED A, GEORGIOU G A, et al. A square root inverse filter for efficient vision-aided inertial navigation on mobile devices[C]//Proceedings of the Robotics: Science and Systems XI, Rome, Jul 13-17, 2015. Cambridge: MIT Press, 2015: 1-9.
[10] BL?SCH M, OMARI S, HUTTER M, et al. Robust visual inertial odometry using a direct EKF-based approach[C]// Proceedings of the 2015 IEEE/RSJ International Conference on Intelligent Robots & Systems, Hamburg, Sep 28-Oct 2, 2015. Piscataway: IEEE, 2015: 298-304.
[11] STRASDAT H, MONTIEL J M M, DAVISON A J. Real-time monocular SLAM: why filter?[C]//Proceedings of the 2010 International Conference on Robotics & Automation, Anchorage, May 3-7, 2010. Piscataway: IEEE, 2010: 2657-2664.
[12] HESCH J A, KOTTAS D G, BOWMAN S L, et al. Con-sistency analysis and improvement of vision-aided inertial navigation[J]. IEEE Transactions on Robotics, 2017, 30(1): 158-176.
[13] SHEN S J, MICHAEL N, KUMAR V. Tightly-coupled monocular visual-inertial fusion for autonomous flight of rotorcraft MAVs[C]//Proceedings of the 2015 IEEE Inter-national Conference on Robotics & Automation, Seattle, May 26-30, 2015. Piscataway: IEEE, 2015: 5303-5310.
[14] YANG Z F, SHEN S J. Monocular visual-inertial state esti-mation with online initialization and camera-IMU extrinsic calibration[J]. IEEE Transactions on Automation Science & Engineering, 2017, 14(1): 39-51.
[15] GUERRA E, MUNGUIA R, GRAU A. Monocular SLAM for autonomous robots with enhanced features Initialization[J]. Sensors, 2014, 14(4): 6317-6337.
[16] KNEIP L, WEISS S, SIEGWART R. Deterministic initiali-zation of metric state estimation filters for loosely-coupled monocular vision-inertial systems[C]//Proceedings of the 2011 IEEE/RSJ International Conference on Intelligent Robots & Systems, San Francisco, Sep 25-30, 2011. Pisca-taway: IEEE, 2011: 2235-2241.
[17] WEISS S, BROCKERS R, ALBREKTSEN S, et al. Inertial optical flow for throw-and-go micro air vehicles[C]//Proc-eedings of the 2015 IEEE Winter Conference on Applications of Computer Vision, Waikoloa, Jan 5-9, 2015. Washington: IEEE Computer Society, 2015: 262-269.
[18] FAESSLER M, FONTANA F, FORSTER C, et al. Automatic re-initialization and failure recovery for aggressive flight with a monocular vision-based quadrotor[C]//Proceedings of the 2015 IEEE International Conference on Robotics & Automation, Seattle, May 26-30, 2015. Piscataway: IEEE, 2015: 1722-1729.
[19] YANG Z, SHEN S. Monocular visual-inertial fusion with online initialization and camera-IMU calibration[C]//Pro-ceedings of the 2015 IEEE International Symposium on Safety, Security, and Rescue Robotics. Piscataway: IEEE, 2015.
[20] MARTINELLI A. Closed-form solution of visual-inertial structure from motion[J]. International Journal of Computer Vision, 2014, 106(2): 138-152.
[21] LI P L, QIN T, HU B T, et al. Monocular visual-inertial state estimation for mobile augmented reality[C]//Proceedings of the 2017 IEEE International Symposium on Mixed and Augmented Reality, Nantes, Oct 9-13, 2017. Washington:IEEE Computer Society, 2017: 11-21.
[22] MUR-ARTAL R, TARDóS J D. Visual-inertial monocular SLAM with map reuse[J]. IEEE Robotics & Automation Letters, 2016, 2(2): 796-803.
[23] FORSTER C, CARLONE L, DELLAERT F, et al. On-manifold preintegration for real-time visual-inertial odo-metry[J]. IEEE Transactions on Robotics, 2017, 33(1): 1-21.
[24] BERCZI L P, BARFOOT T D. State estimation for robotics[M]. New York: Cambridge University Press, 2017.
[25] CHOU J C K. Quaternion kinematic and dynamic differen-tial equations[J]. IEEE Transactions on Robotics and Auto-mation, 1992, 8(1): 53-64.
[26] LUPTON T, SUKKARIEH S. Visual-inertial-aided naviga-tion for high-dynamic motion in built environments without initial conditions[J]. IEEE Transactions on Robotics, 2012, 28(1): 61-76.
[27] CARLONE L, KARAMAN S. Attention and anticipation in fast visual-inertial navigation[C]//Proceedings of the 2017 IEEE International Conference on Robotics and Automa-tion, Singapore, May 29-Jun 3, 2017. Piscataway: IEEE, 2017: 3886-3893.
[28] JOSHI S, BOYD S P. Sensor selection via convex optimi-zation[J]. IEEE Transactions on Signal Processing, 2009, 57(2): 451-462.
[29] BURRI M, NIKOLIC J, GOHL P, et al. The EuRoC micro aerial vehicle datasets[J]. International Journal of Robotics Research, 2016, 35(10): 1157-1163.
[30] AGARWAL S, MIERLE K. Ceres solver[EB/OL]. (2017-11-14) [2020-03-10]. http://ceres-solver.org.