An observability-constrained sliding window filter for SLAM

Huang, Biao; Mourikis,; Roumeliotis,

doi:10.1109/iros.2011.6048760

Cited by 33 publications

(32 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To that extreme, if only the latest sensor state is maintained, we speak of filtering, which amounts the vast majority of related work in VIN [20,21]. Although filtering and fixedlag smoothing enable fast computation, they commit to a linearization point when marginalizing; the gradual build-up of linearization errors leads to drift and possible inconsistencies [22]. A breakthrough in the direction of reconciling filtering and batch optimization has been the development of incremental smoothing techniques (iSAM [23], iSAM2 [24]), which leverage the expressiveness of factor graphs to identify and update only the typically small subset of variables affected by a new measurement.…”

Section: Introductionmentioning

confidence: 99%

IMU Preintegration on Manifold for Efficient Visual-Inertial Maximum-a-Posteriori Estimation

Förster

Carlone

Dellaert

et al. 2015

Robotics: Science and Systems XI

417

368

View full text Add to dashboard Cite

Recent results in monocular visual-inertial navigation (VIN) have shown that optimizationbased approaches outperform filtering methods in terms of accuracy due to their capability to relinearize past states. However, the improvement comes at the cost of increased computational complexity. In this paper, we address this issue by preintegrating inertial measurements between selected keyframes. The preintegration allows us to accurately summarize hundreds of inertial measurements into a single relative motion constraint. Our first contribution is a preintegration theory that properly addresses the manifold structure of the rotation group and carefully deals with uncertainty propagation. The measurements are integrated in a local frame, which eliminates the need to repeat the integration when the linearization point changes while leaving the opportunity for belated bias corrections. The second contribution is to show that the preintegrated IMU model can be seamlessly integrated in a visual-inertial pipeline under the unifying framework of factor graphs. This enables the use of a structureless model for visual measurements, further accelerating the computation. The third contribution is an extensive evaluation of our monocular VIN pipeline: experimental results confirm that our system is very fast and demonstrates superior accuracy with respect to competitive state-of-the-art filtering and optimization algorithms, including off-the-shelf systems such as Google Tango. Abstract-Recent results in monocular visual-inertial navigation (VIN) have shown that optimization-based approaches outperform filtering methods in terms of accuracy due to their capability to relinearize past states. However, the improvement comes at the cost of increased computational complexity. In this paper, we address this issue by preintegrating inertial measurements between selected keyframes. The preintegration allows us to accurately summarize hundreds of inertial measurements into a single relative motion constraint. Our first contribution is a preintegration theory that properly addresses the manifold structure of the rotation group and carefully deals with uncertainty propagation. The measurements are integrated in a local frame, which eliminates the need to repeat the integration when the linearization point changes while leaving the opportunity for belated bias corrections. The second contribution is to show that the preintegrated IMU model can be seamlessly integrated in a visual-inertial pipeline under the unifying framework of factor graphs. This enables the use of a structureless model for visual measurements, further accelerating the computation. The third contribution is an extensive evaluation of our monocular VIN pipeline: experimental results confirm that our system is very fast and demonstrates superior accuracy with respect to competitive state-of-the-art filtering and optimization algorithms, including off-the-shelf systems such as Google Tango [1].

show abstract

Section: Introductionmentioning

confidence: 99%

IMU Preintegration on Manifold for Efficient Visual-Inertial Maximum-a-Posteriori Estimation

Förster

Carlone

Dellaert

et al. 2015

Robotics: Science and Systems XI

417

368

View full text Add to dashboard Cite

show abstract

“…In literature, this initialization is often guessed or assumed to be known [3][4][5][6]. Recently, this sensor fusion problem has been successfully addressed by enforcing observability constraints [7,8] and by using optimization-based approaches [9][10][11][12][13][14][15]. These optimization methods outperform filter-based algorithms in terms of accuracy due to their capability of relinearizing past states.…”

Section: Related Workmentioning

confidence: 99%

Simultaneous State Initialization and Gyroscope Bias Calibration in Visual Inertial Aided Navigation

Kaiser

Martinelli

Fontana

et al. 2017

IEEE Robot. Autom. Lett.

103

View full text Add to dashboard Cite

State of the art approaches for visual-inertial sensor fusion use filter-based or optimizationbased algorithms. Due to the nonlinearity of the system, a poor initialization can have a dramatic impact on the performance of these estimation methods. Recently, a closed-form solution providing such an initialization was derived in [1]. That solution determines the velocity (angular and linear) of a monocular camera in metric units by only using inertial measurements and image features acquired in a short time interval. In this letter, we study the impact of noisy sensors on the performance of this closed-form solution. We show that the gyroscope bias, not accounted for in [1], significantly affects the performance of the method. Therefore, we introduce a new method to automatically estimate this bias. Compared to the original method, the new approach now models the gyroscope bias and is robust to it. The performance of the proposed approach is successfully demonstrated on real data from a quadrotor MAV. Abstract-State of the art approaches for visual-inertial sensor fusion use filter-based or optimization-based algorithms. Due to the nonlinearity of the system, a poor initialization can have a dramatic impact on the performance of these estimation methods. Recently, a closed-form solution providing such an initialization was derived in [1]. That solution determines the velocity (angular and linear) of a monocular camera in metric units by only using inertial measurements and image features acquired in a short time interval. In this paper, we study the impact of noisy sensors on the performance of this closed-form solution. We show that the gyroscope bias, not accounted for in [1], significantly affects the performance of the method. Therefore, we introduce a new method to automatically estimate this bias. Compared to the original method, the new approach now models the gyroscope bias and is robust to it. The performance of the proposed approach is successfully demonstrated on real data from a quadrotor MAV.

show abstract

“…Specifically, sliding-window filters (SWFs) [25], [26] compute a solution for a constantsize, sliding window of states (robot poses and landmark positions) using only the measurements corresponding to that time interval. Similarly, keyframe-based approaches [27]- [29] perform batch optimization over only a (heuristically) selected subset of views or keyframes.…”

Section: A Graph Optimizationmentioning

confidence: 99%

Consistent sparsification for graph optimization

Huang

Kaess

Leonard

2013

2013 European Conference on Mobile Robots

View full text Add to dashboard Cite

Abstract-In a standard pose-graph formulation of simultaneous localization and mapping (SLAM), due to the continuously increasing numbers of nodes (states) and edges (measurements), the graph may grow prohibitively too large for long-term navigation. This motivates us to systematically reduce the pose graph amenable to available processing and memory resources. In particular, in this paper we introduce a consistent graph sparsification scheme: i) sparsifying nodes via marginalization of old nodes, while retaining all the information (consistent relative constraints) -which is conveyed in the discarded measurements -about the remaining nodes after marginalization; and ii) sparsifying edges by formulating and solving a consistent ℓ1-regularized minimization problem, which automatically promotes the sparsity of the graph. The proposed approach is validated on both synthetic and real data.

show abstract

An observability-constrained sliding window filter for SLAM

Cited by 33 publications

References 3 publications

IMU Preintegration on Manifold for Efficient Visual-Inertial Maximum-a-Posteriori Estimation

IMU Preintegration on Manifold for Efficient Visual-Inertial Maximum-a-Posteriori Estimation

Simultaneous State Initialization and Gyroscope Bias Calibration in Visual Inertial Aided Navigation

Consistent sparsification for graph optimization

Contact Info

Product

Resources

About