On-Manifold Preintegration for Real-Time Visual--Inertial Odometry
Open accessThe DOI matches its Crossref record; 83 of the 83 references it cites were found in a registry or an index.
Abstract
Current approaches for visual-inertial odometry (VIO) are able to attain highly accurate state estimation via nonlinear optimization. However, real-time optimization quickly becomes infeasible as the trajectory grows over time; this problem is further emphasized by the fact that inertial measurements come at high rate, hence, leading to the fast growth of the number of variables in the optimization. In this paper, we address this issue by preintegrating inertial measurements between selected keyframes into single relative motion constraints. Our first contribution is a preintegration theory that properly addresses the manifold structure of the rotation group. We formally discuss the generative measurement model as well as the nature of the rotation noise and derive the expression for the maximum a posteriori state estimator. Our theoretical development enables the computation of all necessary Jacobians for the optimization and a posteriori bias correction in analytic form. The second contribution is to show that the preintegrated inertial measurement unit model can be seamlessly integrated into a visual-inertial pipeline under the unifying framework of factor graphs. This enables the application of incremental-smoothing algorithms and the use of a structureless model for visual measurements, which avoids optimizing over the 3-D points, further accelerating the computation. We perform an extensive evaluation of our monocular VIO pipeline on real and simulated datasets. The results confirm that our modeling effort leads to an accurate state estimation in real time, outperforming state-of-the-art approaches.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Multiple View Geometry in Computer Vision
Hartley, R., & Zisserman, A. (2004) · doi:10.1017/cbo9780511811685 · Cited by 21,070
Are we ready for autonomous driving? The KITTI vision benchmark suite
Geiger, A., Lenz, P., & Urtasun, R. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6248074 · Cited by 14,897 · cited 2×
“We use the relative error metrics proposed in [68] to obtain error statistics.”
MethodEstimation with Applications to Tracking and Navigation: Theory, Algorithms and Software
Bar‐Shalom, Y., Li, X. ‐., & Kirubarajan, T. (2002) · doi:10.1002/0471221279 · Cited by 6,782
A Mathematical Introduction to Robotic Manipulation
Murray, R. M., Li, Z., & Sastry, S. S. (2017) · CRC Press eBooks · doi:10.1201/9781315136370 · Cited by 6,694 · cited 1×
“For this purpose, we introduce the following kinematic model [49], [53]:”
MethodFactor graphs and the sum-product algorithm
Kschischang, F. R., Frey, B. J., & Loeliger, H.-A. (2001) · IEEE Transactions on Information Theory · doi:10.1109/18.910572 · Cited by 6,486 · cited 1×
“This naturally leads to the well known factor graph representation, a class of bipartite graphical models that can be used to represent such factored densities [51], [52].”
BackgroundParallel Tracking and Mapping for Small AR Workspaces
Klein, G., & Murray, D. (2007) · 2007 6th IEEE and ACM International Symposium on Mixed and Augmented Reality · doi:10.1109/ismar.2007.4538852 · Cited by 4,346 · cited 1×
“A breakthrough has been the development of incremental smoothing techniques (iSAM [36], iSAM2 [3]), which leverage the expressiveness of factor graphs to maintain sparsity and to identify and update only the typically small subset of variables affected by a new measurement.”
MethodMonoSLAM: Real-time single camera SLAM
Davison, A. J., Reid, I. D., Molton, N. D., & Stasse, O. (2007) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2007.1049 · Cited by 3,959 · cited 1×
“The complexity of the EKF grows quadratically in the number of estimated landmarks, therefore, a small number of landmarks (in the order of 20) are typically tracked to allow real-time operation [10–12].”
BackgroundMatrix Analysis and Applied Linear Algebra
Meyer, C. (2000) · doi:10.1137/1.9780898719512 · Cited by 3,565 · cited 1×
“Any basis El ∈ R2nl×2nl−3 of the null space of El satisfies the following relation [71]:”
BackgroundMatrix analysis and applied linear algebra
(2001) · Choice Reviews Online · doi:10.5860/choice.38-5065 · Cited by 3,300
Stochastic models, estimation, and control
(1982) · Mathematics in Science and Engineering · doi:10.1016/s0076-5392(08)x6186-7 · Cited by 3,229
A Mathmematical Introduction to Robotic Manipulation
Murray, R. M., Li, Z., & Sastry, S. (1994) · Cited by 2,676
SVO: Fast semi-direct monocular visual odometry
Forster, C., Pizzoli, M., & Scaramuzza, D. (2014) · 2014 IEEE International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra.2014.6906584 · Cited by 2,205 · cited 7×
“3http://github.com/uzh-rpg/rpg svo 4http://borg.cc.gatech.edu 14 SVO [65] is a precise and robust monocular visual odometry system that employs sparse image alignment to estimate incremental motion and tracks features by minimizing the photometric error between subsequent frames.”
BackgroundMethodInfluentialSelf-citationA Multi-State Constraint Kalman Filter for Vision-aided Inertial Navigation
Mourikis, A. I., & Roumeliotis, S. I. (2007) · Proceedings 2007 IEEE International Conference on Robotics and Automation · doi:10.1109/robot.2007.364024 · Cited by 1,951 · cited 2×
“The struc-tureless approach requires to keep previous poses in the state vector, by means of stochastic cloning [13].”
BackgroundMethodKeyframe-based visual–inertial odometry using nonlinear optimization
Leutenegger, S., Lynen, S., Bosse, M., Siegwart, R., & Furgale, P. (2015) · The International Journal of Robotics Research · doi:10.1177/0278364914554813 · Cited by 1,723
Stochastic models, estimation, and control
(1979) · Mathematics in Science and Engineering · doi:10.1016/s0076-5392(08)x6125-9 · Cited by 1,624
iSAM2: Incremental smoothing and mapping using the Bayes tree
Kaess, M., Johannsson, H., Roberts, R., Ila, V., Leonard, J. J., & Dellaert, F. (2012) · The International Journal of Robotics Research · doi:10.1177/0278364911430419 · Cited by 1,362 · cited 4×
“nd enables fast inference in real-world problems. 1) Implementation: Our implementation consists of a high frame rate tracking front-end based on SVO3 [65] and an optimization back-end based on iSAM2 [3]4. The front-end tracks salient features in the image at camera rate while the back-end optimizes in parallel the st…”
BackgroundMethodInfluentialSelf-citationiSAM: Incremental Smoothing and Mapping
Kaess, M., Ranganathan, A., & Dellaert, F. (2008) · IEEE Transactions on Robotics · doi:10.1109/tro.2008.2006706 · Cited by 1,076 · cited 1×
“A breakthrough has been the development of incremental smoothing techniques (iSAM [36], iSAM2 [3]), which leverage the expressiveness of factor graphs to maintain sparsity and to identify and update only the typically small subset of variables affected by a new measurement.”
MethodSelf-citationFactor Graphs and GTSAM: A Hands-on Introduction
Dellaert, F. (2012) · SMARTech Repository (Georgia Institute of Technology) · Cited by 972
Self-citationSquare Root SAM: Simultaneous Localization and Mapping via Square Root Information Smoothing
Dellaert, F., & Kaess, M. (2006) · The International Journal of Robotics Research · doi:10.1177/0278364906072768 · Cited by 922 · cited 1×
“(24)”
Self-citationRobust visual inertial odometry using a direct EKF-based approach
Bloesch, M., Omari, S., Hutter, M., & Siegwart, R. (2015) · 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2015.7353389 · Cited by 917 · cited 1×
“The complexity of the EKF grows quadratically in the number of estimated landmarks, therefore, a small number of landmarks (in the order of 20) are typically tracked to allow real-time operation [10–12].”
BackgroundUnified temporal and spatial calibration for multi-sensor systems
Furgale, P., Rehder, J., & Siegwart, R. (2013) · 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems · doi:10.1109/iros.2013.6696514 · Cited by 807 · cited 2×
“1We calibrate the IMU-camera delay using the Kalibr toolbox [59].”
MethodAided Navigation: GPS with High Rate Sensors
Farrell, J. A. (2008) · Cited by 776
Robot sensor calibration: solving AX=XB on the Euclidean group
Park, F. C., & Martin, B. J. (1994) · IEEE Transactions on Robotics and Automation · doi:10.1109/70.326576 · Cited by 570 · cited 1×
“Ground truth poses are provided by a Vicon system mounted in the room; the hand-eye calibration between the Vicon markers and the camera is computed using a least-squares method [67].”
MethodParallel Tracking and Mapping on a camera phone
Klein, G., & Murray, D. (2009) · 2009 8th IEEE International Symposium on Mixed and Augmented Reality · doi:10.1109/ismar.2009.5336495 · Cited by 522 · cited 1×
“Therefore, it has been proposed to discard frames except selected keyframes [24, 32–34] or to run the optimization in a parallel thread, using a tracking”
MethodThe iterated Kalman filter update as a Gauss-Newton method
Bell, B. M., & Cathey, F. W. (1993) · IEEE Transactions on Automatic Control · doi:10.1109/9.250476 · Cited by 506 · cited 1×
“For instance, it can be shown that the iterated EKF equations are equivalent to the Gauss–Newton algorithm, commonly used for smoothing [9].”
MethodIMU Preintegration on Manifold for Efficient Visual-Inertial Maximum-a-Posteriori Estimation
Forster, C., Carlone, L., Dellaert, F., & Scaramuzza, D. (2015) · Robotics: Science and Systems XI · doi:10.15607/rss.2015.xi.006 · Cited by 499
Self-citationReal-time monocular SLAM: Why filter?
Strasdat, H., Montiel, J. M. M., & Davison, A. J. (2010) · 2010 IEEE International Conference on Robotics and Automation · doi:10.1109/robot.2010.5509636 · Cited by 475 · cited 1×
“The front-end also selects a subset of images, called keyframes [32], for which we want to compute a pose estimate.”
BackgroundLupton, T., & Sukkarieh, S. (2012) · IEEE Transactions on Robotics · doi:10.1109/tro.2011.2170332 · Cited by 469 · cited 5×
“An overview of our approach is given in Section IV. The first step towards this goal is the development of a novel preintegration theory. The use of preintegrated IMU measurements was first proposed in [2] and consists of combining many inertial measurements between two keyframes into a single relative motion constraint…”
BackgroundMethodInfluentialTrust-Region Methods on Riemannian Manifolds
Absil, P.-A., Baker, C. G., & Gallivan, K. A. (2007) · Foundations of Computational Mathematics · doi:10.1007/s10208-005-0179-9 · Cited by 461 · cited 4×
“Section III-C reviews optimization on manifolds, following standard references [45].”
BackgroundInfluentialOptimization algorithms exploiting unitary constraints
Manton, J. H. (2002) · IEEE Transactions on Signal Processing · doi:10.1109/78.984753 · Cited by 457 · cited 1×
“Moreover, it provides a grounded and unifying generalization of the error state model, commonly used in aerospace literature for filtering [49] and recently adopted in robotics for optimization [23, 34].”
MethodStochastic models, estimation and control, Volume 1
Maybeck, P. S. (1979) · Cited by 452
Keyframe-Based Visual-Inertial SLAM using Nonlinear Optimization
Leutenegger, S., Furgale, P., Rabaud, V., Chli, M., Konolige, K., & Siegwart, R. (2013) · Robotics: Science and Systems IX · doi:10.15607/rss.2013.ix.037 · Cited by 447
Visual-inertial navigation, mapping and localization: A scalable real-time causal approach
Jones, E. S., & Soatto, S. (2011) · The International Journal of Robotics Research · doi:10.1177/0278364910388963 · Cited by 435 · cited 1×
“The complexity of the EKF grows quadratically in the number of estimated landmarks, therefore, a small number of landmarks (in the order of 20) are typically tracked to allow real-time operation [10]–[12].”
BackgroundRobust Visual Inertial Odometry Using a Direct EKF-Based Approach
Bloesch, M., Omari, S., Hutter, M., et al. (2015) · Repository for Publications and Research Data (ETH Zurich) · doi:10.3929/ethz-a-010566547 · Cited by 424
Means and Averaging in the Group of Rotations
Moakher, M. (2002) · SIAM Journal on Matrix Analysis and Applications · doi:10.1137/s0895479801383877 · Cited by 423
Sigma-point Kalman filtering for integrated GPS and inertial navigation
Crassidis, J. L. (2006) · IEEE Transactions on Aerospace and Electronic Systems · doi:10.1109/taes.2006.1642588 · Cited by 358
Numerical integration of ordinary differential equations on manifolds
Crouch, P. E., & Grossman, R. (1993) · Journal of Nonlinear Science · doi:10.1007/bf02429858 · Cited by 334 · cited 1×
“While we show that this integration scheme performs very well in practice, we remark that for slower IMU measurement rates, one may consider using higher order numerical integration methods [54]–[57].”
Associating Uncertainty With Three-Dimensional Poses for Use in Estimation Problems
Barfoot, T. D., & Furgale, P. T. (2014) · IEEE Transactions on Robotics · doi:10.1109/tro.2014.2298059 · Cited by 305 · cited 2×
“in Section III-C. B. Uncertainty Description in SO(3) A natural definition of uncertainty in SO(3) is to define a distribution in the tangent space, and then map it to SO(3) via the exponential map (6) [44, 46, 47]: ~R = R Exp(); ˘N(0;) ; (12) where R is a given noise-free rotation (the mean) and is a small normally dis…”
BackgroundMartinelli, A. (2012) · IEEE Transactions on Robotics · doi:10.1109/tro.2011.2160468 · Cited by 279 · cited 2×
“A thorough analysis of VIO observability properties is given in [1, 15, 19].”
BackgroundHigh order Runge-Kutta methods on manifolds
Munthe-Kaas, H. (1999) · Applied Numerical Mathematics · doi:10.1016/s0168-9274(98)00030-0 · Cited by 273
Information fusion in navigation systems via factor graph based incremental smoothing
Indelman, V., Williams, S., Kaess, M., & Dellaert, F. (2013) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2013.05.001 · Cited by 271 · cited 1×
“Therefore, inertial measurements are typically integrated between frames to form relative motion constraints [24, 30, 38–40].”
BackgroundSelf-citationA synchronized visual-inertial sensor system with FPGA pre-processing for accurate real-time SLAM
Nikolic, J., Rehder, J., Burri, M., Gohl, P., Leutenegger, S., Furgale, P. T., & Siegwart, R. (2014) · 2014 IEEE International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra.2014.6906892 · Cited by 270 · cited 3×
“with a forward-looking VI-Sensor [66] that consists of an ADIS16448 MEMS IMU and two embedded WVGA monochrome cameras (we only use the left camera).”
MethodInfluentialSliding window filter with application to planetary landing
Sibley, G., Matthies, L., & Sukhatme, G. (2010) · Journal of Field Robotics · doi:10.1002/rob.20360 · Cited by 259
Optimization techniques on Riemannian manifolds
Smith, S. (1995) · Hamiltonian and Gradient Flows, Algorithms and Control · doi:10.1090/fic/003/09 · Cited by 258
Camera-IMU-based localization: Observability analysis and consistency improvement
Hesch, J. A., Kottas, D. G., Bowman, S. L., & Roumeliotis, S. I. (2014) · The International Journal of Robotics Research · doi:10.1177/0278364913509675 · Cited by 245 · cited 2×
“Furthermore, due to marginalization, fixed-lag smoothers share part of the issues of filtering (consistency, build-up of linearization errors) [18, 22, 26].”
BackgroundOnline temporal calibration for camera–IMU systems: Theory and algorithms
Li, M., & Mourikis, A. I. (2014) · The International Journal of Robotics Research · doi:10.1177/0278364913515286 · Cited by 187 · cited 1×
“An alternative is to add the delay as a state in the estimation process [60].”
BackgroundKeyframe-based Visual-Inertial SLAM using Nonlinear Optimization
Leutenegger, S., Furgale, P. T., Rabaud, V., Chli, M., Konolige, K., & Siegwart, R. (2013) · doi:10.3929/ethz-b-000236658 · Cited by 169
Stochastic cloning: a generalized framework for processing relative state measurements
Roumeliotis, S. I., & Burdick, J. W. (2003) · Proceedings 2002 IEEE International Conference on Robotics and Automation (Cat. No.02CH37292) · doi:10.1109/robot.2002.1014801 · Cited by 165 · cited 1×
“A drawback of using a structureless approach for filtering, is that the processing of landmark measurements needs to be delayed until all measurements of a landmark are obtained [5].”
BackgroundMotion Estimation from Image and Inertial Measurements
Strelow, D., & Singh, S. (2004) · The International Journal of Robotics Research · doi:10.1177/0278364904045593 · Cited by 158 · cited 1×
“Full Smoothing Full smoothing methods estimate the entire history of the states (camera trajectory and 3D landmarks), by solving a large nonlinear optimization problem [27–31].”
MethodGeometric integration on Euclidean group with application to articulated multibody systems
Park, J.-H., & Chung, W. K. (2005) · IEEE Transactions on Robotics · doi:10.1109/tro.2005.852253 · Cited by 149 · cited 1×
“While we show that this integration scheme performs very well in practice, we remark that for slower IMU measurement rates one may consider using higher-order numerical integration methods [54–57].”
BackgroundA Square Root Inverse Filter for Efficient Vision-aided Inertial Navigation on Mobile Devices
Wu, K., Ahmed, A., Georgiou, G., & Roumeliotis, S. I. (2015) · Robotics: Science and Systems · doi:10.15607/rss.2015.xi.008 · Cited by 146 · cited 1×
“…for the measurements and the Gaussian priors: the Extended Kalman Filter (EKF) represents the uncertainty using a covariance matrix; instead, information filters and smoothers resort to the information matrix (the inverse of the covariance) or the square-root of the information matrix [3, 8].”
MethodError propagation on the Euclidean group with applications to manipulator kinematics
Wang, Y., & Chirikjian, G. S. (2006) · IEEE Transactions on Robotics · doi:10.1109/tro.2006.878978 · Cited by 136 · cited 1×
“A natural definition of uncertainty in SO(3) is to define a distribution in the tangent space, and then map it to SO(3) via the exponential map (6) [44], [46], [47]:”
BackgroundFactor graph based incremental smoothing in inertial navigation systems
Indelman, V., Williams, S. R., Kaess, M., & Dellaert, F. (2012) · SMARTech Repository (Georgia Institute of Technology) · Cited by 135
Self-citationNonparametric Second-order Theory of Error Propagation on Motion Groups
Wang, Y., & Chirikjian, G. S. (2008) · The International Journal of Robotics Research · doi:10.1177/0278364908097583 · Cited by 131 · cited 2×
“A natural definition of uncertainty in SO(3) is to define a distribution in the tangent space, and then map it to SO(3) via the exponential map (6) [44, 46, 47]:”
BackgroundA First-Estimates Jacobian EKF for Improving SLAM Consistency
Huang, G. P., Mourikis, A. I., & Roumeliotis, S. I. (2009) · Springer tracts in advanced robotics · doi:10.1007/978-3-642-00196-3_43 · Cited by 126 · cited 1×
“This problem was addressed with the first-estimates jacobian approach [17], which ensures that a state is not updated with different linearization points — a source of inconsistency.”
Dellaert, F. (2005) · Robotics: Science and Systems · doi:10.15607/rss.2005.i.024 · Cited by 123
Self-citationMotion tracking with fixed-lag smoothing: Algorithm and consistency analysis
Dong-Si, T.-C., & Mourikis, A. I. (2011) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2011.5980267 · Cited by 123
Carlone, L., Kira, Z., Beall, C., Indelman, V., & Dellaert, F. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6907483 · Cited by 116 · cited 1×
“The use of preintegrated IMU measurements was first proposed in [2] and consists of combining many inertial measurements between two keyframes into a single relative motion constraint.”
MethodSelf-citationA Spline-Based Trajectory Representation for Sensor Fusion and Rolling Shutter Cameras
Patron-Perez, A., Lovegrove, S., & Sibley, G. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0811-3 · Cited by 104 · cited 1×
“Full smoothing methods estimate the entire history of the states (camera trajectory and 3D landmarks), by solving a large nonlinear optimization problem [27–31].”
MethodStochastic models, information theory, and Lie groups
Chirikjian, G. S. (2009) · Cited by 96
An observability-constrained sliding window filter for SLAM
Huang, G. P., Mourikis, A. I., & Roumeliotis, S. I. (2011) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2011.6095161 · Cited by 88 · cited 1×
“Furthermore, due to marginalization, fixed-lag smoothers share part of the issues of filtering (consistency, build-up of linearization errors) [18, 22, 26].”
BackgroundOn the Consistency of Vision-Aided Inertial Navigation
Kottas, D. G., Hesch, J. A., Bowman, S. L., & Roumeliotis, S. I. (2013) · Springer Tracts in Advanced Robotics · doi:10.1007/978-3-319-00065-7_22 · Cited by 76 · cited 2×
“A thorough analysis of VIO observability properties is given in [1, 15, 19].”
BackgroundA dual-layer estimator architecture for long-term localization
Mourikis, A. I., & Roumeliotis, S. I. (2008) · 2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops · doi:10.1109/cvprw.2008.4563131 · Cited by 67 · cited 2×
“Therefore, it has been proposed to discard frames except selected keyframes [24, 32–34] or to run the optimization in a parallel thread, using a tracking 4 and mapping dual architecture [20, 35].”
BackgroundMethodFactor graph based incremental smoothing in inertial navigation systems
Indelman, Williams, Kaess, & Dellaert (2012) · Cited by 66
C-KLAM: Constrained keyframe-based localization and mapping
Nerurkar, E. D., Wu, K. J., & Roumeliotis, S. I. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6907385 · Cited by 65 · cited 2×
“Therefore, it has been proposed to discard frames except selected keyframes [24], [32]–[34] or to run the optimization in a parallel thread, using a tracking and mapping dual architecture [20], [35].”
MethodAirborne smoothing and mapping using vision and inertial sensors
Bryson, M., Johnson‐Roberson, M., & Sukkarieh, S. (2009) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/robot.2009.5152678 · Cited by 63 · cited 1×
“Full smoothing guarantees the highest accuracy; however, real-time operation quickly becomes infeasible as the trajectory and the map grow over time.”
BackgroundGeometric Integration of Quaternions
Andrle, M. S., & Crassidis, J. L. (2013) · Journal of Guidance Control and Dynamics · doi:10.2514/1.58558 · Cited by 61 · cited 1×
“While we show that this integration scheme performs very well in practice, we remark that for slower IMU measurement rates, one may consider using higher order numerical integration methods [54]–[57].”
Factor graphs and the sum–product algorithm
Wymeersch, H. (2007) · Cambridge University Press eBooks · doi:10.1017/cbo9780511619199.006 · Cited by 59
Camera trajectory estimation using inertial sensor measurements and structure from motion results
Jung, S., & Taylor, C. J. (2005) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2001.991037 · Cited by 49 · cited 1×
“Full Smoothing Full smoothing methods estimate the entire history of the states (camera trajectory and 3D landmarks), by solving a large nonlinear optimization problem [27–31].”
MethodObservability, identifiability and sensitivity of vision-aided inertial navigation
Hernandez, J., Tsotsos, K., & Soatto, S. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139507 · Cited by 38 · cited 1×
“A thorough analysis of VIO observability properties is given in [1, 15, 19].”
Robust inference for visual-inertial sensor fusion
Tsotsos, K., Chiuso, A., & Soatto, S. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139924 · Cited by 31 · cited 1×
“Further, linearization errors introduce drift in the estimate and render the filter inconsistent.”
BackgroundRepresentation issues in the ML estimation of camera motion
Joachim Hornegger, A., & Tomasi, C. (1999) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.1999.791285 · Cited by 31 · cited 1×
“Moreover, Euler angles are known to have singularities.”
BackgroundIncremental light bundle adjustment for robotics navigation
Indelman, V., Melim, A., & Dellaert, F. (2013) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2013.6696615 · Cited by 29 · cited 1×
“For standard IMU integration between two frames, the initial condition is given by the state estimate at the first frame.”
BackgroundSelf-citationAn observability-constrained sliding window filter for SLAM
Huang, G. P., Mourikis, A. I., & Roumeliotis, S. I. (2011) · 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems · doi:10.1109/iros.2011.6048760 · Cited by 28
Asynchronous Adaptive Conditioning for Visual-Inertial SLAM
Keivan, N., Patron-Perez, A., & Sibley, G. (2015) · Springer tracts in advanced robotics · doi:10.1007/978-3-319-23778-7_21 · Cited by 27 · cited 1×
“For standard IMU integration between two frames, the initial condition is given by the state estimate at the first frame.”
BackgroundAutonomous Navigation in Complex Indoor and Outdoor Environments with Micro Aerial Vehicles
Shen, S. (2014) · ScholarlyCommons (University of Pennsylvania) · Cited by 26
Observability Properties and Deterministic Algorithms in Visual-Inertial Structure from Motion
Martinelli, A. (2012) · Foundations and Trends in Robotics · doi:10.1561/2300000030 · Cited by 24 · cited 2×
“A thorough analysis of VIO observability properties is given in [1, 15, 19].”
BackgroundFörster, C., Carlone, L., Dellaert, F., & Scaramuzza, D. (2015) · SMARTech Repository (Georgia Institute of Technology) · Cited by 17
Self-citationThe Bayes Tree: Enabling Incremental Reordering and Fluid Relinearization for Online Mapping
Kaess, M., Ila, V., Roberts, R. B., & Dellaert, F. (2010) · DSpace@MIT (Massachusetts Institute of Technology) · Cited by 7
Self-citationGeometric Integration of Quaternions
Andrle, M. S., & Crassidis, J. L. (2012) · AIAA/AAS Astrodynamics Specialist Conference and Exhibit · doi:10.2514/6.2012-4421 · Cited by 7
Statistical modeling of relations for 3-D object recognition
Hornegger, J. (2002) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.1997.595466 · Cited by 4 · cited 1×
“The second drawback is that the Euler parametrization is not fair [64], which means that, given the preintegrated Euler angles θ̃, the negative log-likelihood L(θ) = (1)2 ‖θ̃ − θ‖(2)Σ is not invariant under the action of rigid body transformations.”
BackgroundObservability Properties and Deterministic Algorithms in Visual-Inertial Structure from Motion
Martinelli, A. (2012) · now publishers, Inc. eBooks · doi:10.1561/9781601987396 · Cited by 3
Camera Trajectory Estimation using Inertial Sensor Measurements and Structure fom Motion Results
Jung, S.-H., & Taylor, C. J. (2001) · ScholarlyCommons (University of Pennsylvania) · Cited by 2
Cited by
VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator
Qin, T., Li, P., & Shen, S. (2018) · IEEE Transactions on Robotics · doi:10.1109/tro.2018.2853729 · Cited by 4,669
“We note that our current IMU preintegration procedure shares almost the same numerical results as [19] and [24], but using different derivations.”
MethodResultInfluentialORB-SLAM3: An Accurate Open-Source Library for Visual, Visual–Inertial, and Multimap SLAM
Campos, C., Elvira, R., Rodriguez, J. J. G., M. Montiel, J. M., & D. Tardos, J. (2021) · IEEE Transactions on Robotics · doi:10.1109/tro.2021.3075644 · Cited by 4,274
Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age
Cadena, C., Carlone, L., Carrillo, H., Latif, Y., Scaramuzza, D., Neira, J., Reid, I., & Leonard, J. J. (2016) · IEEE Transactions on Robotics · doi:10.1109/tro.2016.2624754 · Cited by 3,596
“et of linear equations (the so called normal equations). These approaches can be seamlessly generalized to variables belonging to smooth manifolds (e.g., rotations), which are of interest in robotics [1, 86]. The key insight behind modern SLAM solvers is that the matrix appearing in the normal equations is sparse and…”
Self-citationPast, present, and future of simultaneous localization and mapping: Toward the robust-perception age
Cadena, C., Carlone, L., Carrillo, H., Latif, Y., Scaramuzza, D., Neira, J., Reid, I. D., & Leonard, J. J. (2016) · Zurich Open Repository and Archive (University of Zurich) · doi:10.5167/uzh-138892 · Cited by 3,048
Self-citationLIO-SAM: Tightly-coupled Lidar Inertial Odometry via Smoothing and Mapping
Shan, T., Englot, B., Meyers, D., Wang, W., Ratti, C., & Rus, D. (2020) · 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341176 · Cited by 2,199
“Due to space limitations, we refer the reader to the description from [20] for the detailed derivation of Eqs.”
Method
Metadata
- Authors
- Forster, C., Carlone, L., Dellaert, F., & Scaramuzza, D.
- Affiliations
- Massachusetts Institute of Technology (US)
- Georgia Institute of Technology (US)
- University of Zurich (CH)
- Venue
- IEEE Transactions on Robotics
- Published
- 2017-02
- Volume
- 33
- Issue
- 1
- Pages
- 1-21
- ISSN
- 1552-3098 (ISSN-L)
- 1941-0468
- OpenAlex
- W2482726005
- Cited by
- 1,502
- References
- 83
- Citations by year
- 2026: 156 · 2025: 179 · 2024: 234 · 2023: 193 · 2022: 180 · 2021: 174 · 2020: 134 · 2019: 129 · 2018: 88 · 2017: 32
- FWCI
- 102.35
- Type
- journal article
- Language
- en
- Open access
- green
- www.zora.uzh.ch
- Licence
- Locations
- IEEE Transactions on Robotics (publishedVersion)
- Zurich Open Repository and Archive (University of Zurich) (acceptedVersion, other-oa)
- arXiv (Cornell University) (submittedVersion)
- Infoscience (Ecole Polytechnique Fédérale de Lausanne) (submittedVersion)
- Infoscience (Ecole Polytechnique Fédérale de Lausanne) (submittedVersion, cc-by-nc-nd)
- Zurich Open Repository and Archive (University of Zurich) (submittedVersion, other-oa)
- arXiv (Cornell University)
- Zurich Open Repository and Archive (University of Zurich)
- Funders
- Swiss National Foundation (200021-143607)
- National Center of Competence in Research Robotics (NCCR)
- University of Zurich
- National Science Foundation (11115678)
- United States Department of Agriculture (USDA)
- National Institute of Food and Agriculture (GEOW-2014-09160)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Robotics and Sensor-Based Localization
- Physical Sciences › Engineering › Aerospace Engineering › Inertial Sensor and Navigation
- Physical Sciences › Computer Science › Artificial Intelligence › Target Tracking and Data Fusion in Sensor Networks
- Keywords
- visual-inertial odometry, inertial measurement unit, factor graphs, maximum a posteriori estimation, bias correction, real-time state estimation, keyframe selection
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.