Towards a robust back-end for pose graph SLAM
The DOI matches its Crossref record; 28 of the 33 references it cites were found in a registry or an index.
Abstract
Current state of the art solutions of the SLAM problem are based on efficient sparse optimization techniques and represent the problem as probabilistic constraint graphs. For example in pose graphs the nodes represent poses and the edges between them express spatial information (e.g. obtained from odometry) and information on loop closures. The task of constructing the graph is delegated to a front-end that has access to the available sensor information. The optimizer, the so called back-end of the system, relies heavily on the topological correctness of the graph structure and is not robust against misplaced constraint edges. Especially edges representing false positive loop closures will lead to the divergence of current solvers. We propose a novel formulation that allows the back-end to change parts of the topological structure of the graph during the optimization process. The back-end can thereby discard loop closures and converge towards correct solutions even in the presence of false positive loop closures. This largely increases the overall robustness of the SLAM system and closes a gap between the sensor-driven front-end and the back-end optimizers. We demonstrate the approach and present results both on large scale synthetic and real-world datasets.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
Thrun, S. (2002) · Communications of the ACM · doi:10.1145/504729.504754 · Cited by 7,994 · cited 1×
“, see [30] for an introduction to these methods), optimization-based approaches build upon efficient algorithms for nonlinear optimization that exploit the sparsity inherent in the SLAM problem.”
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×
“The pose graph SLAM problem can be conveniently modelled as a factor graph [17].”
BackgroundRobust Regression: Asymptotics, Conjectures and Monte Carlo
Huber, P. J. (1973) · The Annals of Statistics · doi:10.1214/aos/1176342503 · Cited by 2,328 · cited 1×
“So called robust cost functions, like the Huber function [12], can reduce the influence of potential outliers.”
BackgroundG2o: A general framework for graph optimization
Kummerle, R., Grisetti, G., Strasdat, H., Konolige, K., & Burgard, W. (2011) · 2011 IEEE International Conference on Robotics and Automation · doi:10.1109/icra.2011.5979949 · Cited by 1,982 · cited 6×
“Due to its inherent sparse structure the problem can be solved efficiently with state of the art frameworks like g2o [18] or GTSAM [1] that use iterative solvers such as Gauss-Newton or LevenbergMarquardt.”
BackgroundMethodInfluentialFAB-MAP: Probabilistic Localization and Mapping in the Space of Appearance
Cummins, M., & Newman, P. (2008) · The International Journal of Robotics Research · doi:10.1177/0278364908090961 · Cited by 1,472 · cited 2×
“Furthermore, the front-end part of the system can detect loop closures between two poses xi and xj , for example by a visual place recognition system like [4] or [29].”
BackgroundGlobally Consistent Range Scan Alignment for Environment Mapping
Lu, F., & Milios, E. (1997) · Autonomous Robots · doi:10.1023/a:1008854305733 · Cited by 1,304
iSAM: 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 3×
“The datasets City and Sphere shipped with the open-source implementation of iSAM [14].”
BackgroundMethodSquare 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 10×
“1, which shows the results of iSAM [14] version 1.5. for 10 outlier constraints.”
BackgroundInfluentialData association in stochastic mapping using the joint compatibility test
Neira, J., & Tardos, J. D. (2001) · IEEE Transactions on Robotics and Automation · doi:10.1109/70.976019 · Cited by 772 · cited 1×
“Although many of these strategies [2], [4], [15], [23], [25], [27] are appealing and reach high precision and recall measures in several benchmarks, the bottom line is that none of the current approaches can guarantee to always work perfectly and never let a false positive pass.”
BackgroundEfficient Sparse Pose Adjustment for 2D mapping
Konolige, K., Grisetti, G., Kümmerle, R., Burgard, W., Limketkai, B., & Vincent, R. (2010) · 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems · doi:10.1109/iros.2010.5649043 · Cited by 395 · cited 2×
“{niko.suenderhauf, peter.protzel}@etit.tu-chemnitz.de iSAM [14] and very recently Sparse Pose Adjustment [16], iSAM2 [13], and g2o [18].”
Background
Research works printed but not linked: 2
- ku?mmerle, 0, G2o: A general framework for graph optimization, Proc of the IEEE Int Conf on Robotics and Automation (ICRA) 2011
- olson, 0, Fast iterative optimization of pose graphs with poor initial estimates, Inl Conf on Robotics and Automation ICRA 2006
Cited by
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
“In the colored box below we provide an example of SLAM for a robot equipped with a monocular camera and a 3D laser scanner.”
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. D., & Leonard, J. J. (2016) · Zurich Open Repository and Archive (University of Zurich) · doi:10.5167/uzh-138892 · Cited by 3,048
Online global loop closure detection for large-scale multi-session graph-based SLAM
Labbe, M., & Michaud, F. (2014) · 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems · doi:10.1109/iros.2014.6942926 · Cited by 373
“Some graph optimization approaches such as [22], [23] deal with possible invalid matches, and could be used to increase robustness of the proposed approach.”
MethodSwitchable constraints for robust pose graph SLAM
Sunderhauf, N., & Protzel, P. (2012) · 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems · doi:10.1109/iros.2012.6385590 · Cited by 344
“Notice that the system’s feasibility to work on large-scale real-world datasets has been demonstrated before [14].”
BackgroundMethodInfluentialSelf-citationA Survey on Active Simultaneous Localization and Mapping: State of the Art and New Frontiers
Placed, J. A., Strader, J., Carrillo, H., Atanasov, N., Indelman, V., Carlone, L., & Castellanos, J. A. (2023) · IEEE Transactions on Robotics · doi:10.1109/tro.2023.3248510 · Cited by 332
A review of visual SLAM methods for autonomous driving vehicles
Cheng, J., Zhang, L., Chen, Q., Hu, X., & Cai, J. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.104992 · Cited by 252
Inference on networks of mixtures for robust robot mapping
Olson, E., & Agarwal, P. (2013) · The International Journal of Robotics Research · doi:10.1177/0278364913479413 · Cited by 252
“In the previous sections, we demonstrated that our method can be used to encode null hypotheses, or equivalently, implement robustified cost functions: a capability similar to earlier work (Sunderhauf and Protzel, 2012).”
InfluentialPairwise Consistent Measurement Set Maximization for Robust Multi-Robot Map Merging
Mangelson, J. G., Dominic, D., Eustice, R. M., & Vasudevan, R. (2018) · 2018 IEEE International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra.2018.8460217 · Cited by 202
“Switchable constraints [4] proposes the use of of switchable error factors in the SLAM back-end solver that enable pose constraints with high residual error to be “turned off” so that they no longer affect the solution.”
BackgroundOpenRatSLAM: an open source brain-based SLAM system
Ball, D., Heath, S., Wiles, J., Wyeth, G. F., Corke, P., & Milford, M. J. (2013) · Autonomous Robots · doi:10.1007/s10514-012-9317-9 · Cited by 152
Recent developments and applications of simultaneous localization and mapping in agriculture
Ding, H., Zhang, B., Zhou, J. C., Yan, Y., Tian, G., & Gu, B. (2022) · Journal of Field Robotics · doi:10.1002/rob.22077 · Cited by 146
“To simplify the complex, some groups suggested to ignore the optimization of many feature points (landmarks) and focus only on the trajectory of the keyframe (the edge between poses), that is, the Pose Graph (Dubbelman & Browning, 2015; Sunderhauf & Protzel, 2012).”
BackgroundA fast and accurate approximation for planar pose graph optimization
Carlone, L., Aragues, R., Castellanos, J. A., & Bona, B. (2014) · The International Journal of Robotics Research · doi:10.1177/0278364914523689 · Cited by 101
“These methods include also the topology of the graph in the optimization process, activating or deactivating constraints with suitable switch variables (Sünderhauf and Protzel, 2012a,b), or using a max-mixture model (Olson and Agarwal, 2012).”
MethodAitken, J. M., Evans, M. H., Worley, R., Edwards, S., Zhang, R., Dodd, T. J., Mihaylova, L. S., & Anderson, S. (2021) · IEEE Access · doi:10.1109/access.2021.3115981 · Cited by 116
“This type of approach is used in early work in dynamic covariance scaling [75] and related methods such as switchable constraints [76], [77] and max mixtures [78], which use a tunable weighting to decrease the”
MethodRobust pose-graph loop-closures with expectation-maximization
Lee, G. H., Fraundorfer, F., & Pollefeys, M. (2013) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2013.6696406 · Cited by 92
“The task of identifying erroneous loop-closure constraints is always left to the front-end, and it is only until the recent two years that several works [5]–[9] demonstrated the ability robustly detect and disable erroneous loop-closure constraints within the back-end optimization process.”
BackgroundMethodSLAM with objects using a nonparametric pose graph
Mu, B., Liu, S.-Y., Paull, L., Leonard, J. J., & How, J. P. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759677 · Cited by 89
Simultaneous Localization and Mapping in the Epoch of Semantics: A Survey
Sualeh, M., & Kim, G.-W. (2018) · International Journal of Control Automation and Systems · doi:10.1007/s12555-018-0130-x · Cited by 94
“An incorrect data association makes a false loop closure unavoidable and is fed to the back end [135].”
BackgroundModeling Perceptual Aliasing in SLAM via Discrete–Continuous Graphical Models
Lajoie, P., Hu, S., Beltrame, G., & Carlone, L. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2894852 · Cited by 108
“Sünderhauf and Protzel [7], [24] propose Vertigo, which augments the PGO problem with latent binary variables (then relaxed to continuous variables) that are responsible for deactivating outliers.”
BackgroundCarlone, L., & Censi, A. (2014) · IEEE Transactions on Robotics · doi:10.1109/tro.2013.2291626 · Cited by 79
“Olson and Agarwal [19], and Sünderhauf and Protzel [20], [21] extend this framework, with the purpose of”
BackgroundLong-term online multi-session graph-based SPLAM with memory management
Labbé, M., & Michaud, F. (2017) · Autonomous Robots · doi:10.1007/s10514-017-9682-5 · Cited by 90
“If this happens, graph optimization approaches such as (Latif et al., 2013; Sunderhauf and Protzel, 2012; Lee et al., 2013) deal with possible invalid matches, and could be used to increase robustness of SPLAM-MM.”
MethodSimultaneous Localization And Mapping: Present, Future, and the Robust-Perception Age.Preprint
Cadena, C., Carlone, L., Carrillo, H., Latif, Y., Scaramuzza, D., Neira, J., Reid, I. D., & Leonard, J. J. (2016) · arXiv (Cornell University) · Cited by 90
Modeling Perceptual Aliasing in SLAM via Discrete–Continuous Graphical ModelsPreprint
Lajoie, P., Hu, S., Beltrame, G., & Carlone, L. (2019) · DSpace@MIT (Massachusetts Institute of Technology) · Cited by 76
Indelman, V., Nelson, E., Michael, N., & Dellaert, F. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6906915 · Cited by 77
“[23]), will typically fail as further discussed in Section V.”
BackgroundRobust LiDAR-based localization in architectural floor plans
Boniardi, F., Caselitz, T., Kummerle, R., & Burgard, W. (2017) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2017.8206168 · Cited by 70
“The backbone of the method consists in a maximum a posteriori pose-graph-based SLAM system [18], which”
MethodConvex Relaxations for Pose Graph Optimization With Outliers
Carlone, L., & Calafiore, G. C. (2018) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2018.2793352 · Cited by 84
“S¨underhauf and Protzel, in [30], [31], propose to augment the PGO problem with latent binary variables that are responsible for deactivating outliers.”
BackgroundCOP-SLAM: Closed-Form Online Pose-Chain Optimization for Visual SLAM
Dubbelman, G., & Brett Browning, A. (2015) · IEEE Transactions on Robotics · doi:10.1109/tro.2015.2473455 · Cited by 67
Outlier-Robust Estimation: Hardness, Minimally Tuned Algorithms, and Applications
Antonante, P., Tzoumas, V., Yang, H., & Carlone, L. (2021) · IEEE Transactions on Robotics · doi:10.1109/tro.2021.3094984 · Cited by 78
How to learn an illumination robust image feature for place recognition
Lategahn, H., Beck, J., Kitt, B., & Stiller, C. (2013) · 2013 IEEE Intelligent Vehicles Symposium (IV) · doi:10.1109/ivs.2013.6629483 · Cited by 66
“Thereby we are able to double the number of descriptors that can be evaluated per time and speed up the training accordingly.”
BackgroundMH-iSAM2: Multi-hypothesis iSAM using Bayes Tree and Hypo-tree
Hsiao, M., & Kaess, M. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8793854 · Cited by 65
“There are other back-end solutions that change the topology of the underlying graph [19][20] or self-tuning the destribution parameters [17] during optimization with ambiguity, or add a step before conventional optimizers to choose the best hypothesis [18].”
BackgroundA robust pose graph approach for city scale LiDAR mapping
Yang, S. Y., Zhu, X., Nian, X., Feng, L., Qu, X., & Ma, T. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593754 · Cited by 66
“In the back-end, many works are introduced to reject erroneous factors: Both SC [25] and DCS [26] propose to add a robust kernel for each factor to remove outliers in the optimization progress.”
BackgroundMultipath mitigation in GNSS-based localization using robust optimization
Sünderhauf, N., Obst, M., Wanielik, G., & Protzel, P. (2012) · IEEE Intelligent Vehicles Symposium · doi:10.1109/ivs.2012.6232299 · Cited by 57
“It has been shown that both interpretations are equivalent [2].”
BackgroundMethodSelf-citationSensor fusion for flexible human-portable building-scale mapping
Fallon, M. F., Johannsson, H., Brookshire, J., Teller, S. J., & Leonard, J. J. (2012) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2012.6385882 · Cited by 63
“An interesting approach proposed in [24] achieves a robustness to false loop-closures by disabling those with inordinate optimization cost.”
BackgroundFeature-based visual simultaneous localization and mapping: a survey
Azzam, R., Taha, T., Huang, S., & Zweiri, Y. (2020) · SN Applied Sciences · doi:10.1007/s42452-020-2001-3 · Cited by 63
“Alternatively, successive robot poses in the environment could be used alone to estimate the location of the robot using a Pose Graph [119].”
MethodVertical Corner Feature Based Precise Vehicle Localization Using 3D LIDAR in Urban Area
Im, J.-H., Im, S.-H., & Jee, G.-I. (2016) · Sensors · doi:10.3390/s16081268 · Cited by 55
“T e vehicle trajectory was optimized by using the GraphSLAM method [28,29].”
MethodInfluentialAUV robust bathymetric simultaneous localization and mapping
Ma, T., Li, Y., Wang, R., Zheng, C., & Gong, Y. (2018) · Ocean Engineering · doi:10.1016/j.oceaneng.2018.08.029 · Cited by 56
Time-Relative RTK-GNSS: GNSS Loop Closure in Pose Graph Optimization
Suzuki, T. (2020) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2020.3003861 · Cited by 53
Large-Scale Outdoor SLAM Based on 2D Lidar
Ren, R., Fu, H., & Wu, M. (2019) · Electronics · doi:10.3390/electronics8060613 · Cited by 50
“Some recent researches focused on utilizing these algorithms to reject false loop closures, such as the Max-Mixture Model [28], Switchable Constraints [29], and Realizing, Reversing, Recovering (RRR) [30].”
BackgroundEfficient onbard RGBD-SLAM for autonomous MAVs
Scherer, S., & Zell, A. (2013) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2013.6696482 · Cited by 56
“We use HOG-MAN [17] to optimize the pose graph consisting of keyframes (nodes) and relative poses (edges) that were added after registering pairs of keyframes.”
MethodStructure-based vision-laser matching
Gawel, A., Cieslewski, T., Dubé, R., Bosse, M., Siegwart, R., & Nieto, J. I. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759053 · Cited by 47
“Considering recent advances in robust estimation lower precision rates are commonly acceptable as false positives may be filtered [20], [21].”
BackgroundNovel insights into the impact of graph structure on SLAM
Khosoussi, K., Huang, S., & Dissanayake, G. (2014) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2014.6942932 · Cited by 56
“For example, data association has always been and still is a challenge [4].”
BackgroundA Comprehensive Review of Humanoid Robots
Sheng, Q., Zhou, Z., Li, J., Mi, X., Xiang, P., Chen, Z., Xu, H., Jia, S., Wu, X., Cui, Y., et al. (2025) · SmartBot · doi:10.1002/smb2.12008 · Cited by 70
Simultaneous localization and mapping with multimodal probability distributions
Pfingsthorn, M. F., & Birk, A. (2012) · The International Journal of Robotics Research · doi:10.1177/0278364912461540 · Cited by 45
“Sünderhauf and Protzel (2012) add an interesting extension to the work of both Dellaert (2005) and Kümmerle et al. (2011), namely switch variables that are part of the optimization and can switch off constraints.”
BackgroundOutlier-Robust Spatial Perception: Hardness, General-Purpose Algorithms, and Guarantees
Tzoumas, V., Antonante, P., & Carlone, L. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968174 · Cited by 47
“tlier-robust SLAM has traditionally relied on M-estimators; e.g., [9]. Olson and Agarwal [41] use a max-mixture distribution to approximate multi-modal measurement noise. Sünderhauf and Protzel [40], [55] augment the problem with latent binary variables responsible for deactivating outliers. Tong and Barfoot [56], [57…”
MethodCharacterizing the Uncertainty of Jointly Distributed Poses in the Lie Algebra
Mangelson, J. G., Ghaffari, M., Vasudevan, R., & Eustice, R. M. (2020) · IEEE Transactions on Robotics · doi:10.1109/tro.2020.2994457 · Cited by 63
“However, while recent years have seen an increase in use of Lie-algebra-based methods for uncertainty propagation [30]–[32], existing methods assume that individual measurements are independent [1], which may not be the case when the underlying pose estimates are derived from a SLAM solution.”
MethodInfluentialFactor Graphs for Navigation Applications: A Tutorial
Taylor, C. N., & Gross, J. N. (2024) · NAVIGATION Journal of the Institute of Navigation · doi:10.33012/navi.653 · Cited by 50
Pfingsthorn, M. F., & Birk, A. (2015) · The International Journal of Robotics Research · doi:10.1177/0278364915585395 · Cited by 42
“It is obvious that, even though there is a high variance for all methods, Generalized Prefilter + Cauchy GN performs multiple orders of magnitude better than Max-Mixture GN and DCS GN, and around an order of magnitude better than Switchable Constraints GN.”
BackgroundMethodInfluentialTowards robust graphical models for GNSS-based localization in urban environments
Sünderhauf, N., & Protzel, P. (2012) · International Multi-Conference on Systems, Signals & Devices (SSD) · doi:10.1109/ssd.2012.6198059 · Cited by 39
“In recent previous work [5], [6] we developed an approach for robust optimization for pose graph SLAM problems in the presence of such outliers.”
BackgroundMethodSelf-citationRobust Estimation and Applications in Robotics
Bosse, M. C., Agamennoni, G., & Gilitschenski, I. (2016) · Foundations and Trends in Robotics · doi:10.1561/2300000047 · Cited by 53
Robust incremental SLAM with consistency-checking
Graham, M. C., How, J. P., & Gustafson, D. E. (2015) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2015.7353363 · Cited by 44
“However, if incorrect landmark or loop closure measurements occur in the data, standard NLS solvers will fail to converge to the correct solution [5,6].”
BackgroundMethodInfluentialExtended Line Map-Based Precise Vehicle Localization Using 3D LIDAR
Im, J.-H., Im, S.-H., & Jee, G.-I. (2018) · Sensors · doi:10.3390/s18103179 · Cited by 38
“The theory and principles for the GraphSL ethod are ell described in [22,23].”
MethodFast Reconstruction of 3D Point Cloud Model Using Visual SLAM on Embedded UAV Development Platform
Huang, F., Yang, H., Tan, X., Peng, S., Tao, J., & Peng, S. (2020) · Remote Sensing · doi:10.3390/rs12203308 · Cited by 40
Robust graph SLAM back-ends: A comparative analysis
Latif, Y., Cadena, C., & Neira, J. (2014) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2014.6942929 · Cited by 36
“This work is more focused on real world scenarios in which there are a lot of false positives and very few (or no) true positives.”
BackgroundIndelman, V., Nelson, E., Dong, J., Michael, N., & Dellaert, F. (2016) · IEEE Control Systems · doi:10.1109/mcs.2015.2512031 · Cited by 32
Agamennoni, G., Furgale, P., & Siegwart, R. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139840 · Cited by 40
“We briefly discuss four recent approaches to robustifying the SLAM back-end, namely: max-mixtures [5], [6], switchable constraints [7], [8], [9], dynamic covariance scaling [10], [11] and maximum a posteriori inference [12].”
BackgroundMethodMapping with synthetic 2D LIDAR in 3D urban environment
Chong, Z. J., Qin, B., Bandyopadhyay, T., Ang, M. H., Frazzoli, E., & Rus, D. L. (2013) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2013.6697035 · Cited by 35
“The leading examples include Switchable Constraints [24], Max-Mixture Model [19], and Realizing, Reversing, Recovering (RRR) [13].”
BackgroundEnhanced and Facilitated Indoor Positioning by Visible-Light GraphSLAM Technique
Yue, Y., Zhao, X., & Li, Z. (2020) · IEEE Internet of Things Journal · doi:10.1109/jiot.2020.3012463 · Cited by 32
“We assume that the successive poses are the normally distributed random variables [34] as where s i is obtained from a PDR algorithm to denote the pose change between the i th and i + 1th pose, and (cid”
BackgroundMethodInfluentialSun, L. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3116313 · Cited by 38
“However, local M-estimation methods (e.g. [25], [39]) highly depend on a promising initial guess, thus prone to fail in scenes lacking the initialization information.”
MethodRealizing, reversing, recovering: Incremental robust loop closing over time using the iRRR algorithm
Latif, Y., Cadena, C., & Neira, J. (2012) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2012.6385879 · Cited by 35
“RRR algorithm) against our own implementation of Sunderhauf’s approach to robust back-end [8] using the parameters and functions for switch links as given in their paper.”
BackgroundMethodInfluentialRobust Incremental SLAM Under Constrained Optimization Formulation
Bai, F., Vidal‐Calleja, T., & Huang, S. (2018) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2018.2794610 · Cited by 30
“The first line of work introduces additional variables, either to the optimization problem [18] or to the probabilistic graphical”
BackgroundFu, T., Su, S., Lu, Y., & Wang, C. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3382533 · Cited by 30
Do, H., Hong, S., & Kim, J. (2020) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2020.3010731 · Cited by 32
“Switchable constraints [5], [6] suggests using a new variable to adjust the weights of the loop closures depending on the residual error.”
A hybrid topological mapping and navigation method for large area robot mapping
Ravankar, A. A., Ravankar, A. A., Emaru, T., & Kobayashi, Y. (2017) · Annual Conference of the Society of Instrument and Control Engineers of Japan (SICE) · doi:10.23919/sice.2017.8105770 · Cited by 28
“The solution can be regarded as a least square problem which can be solved using the pose graph SLAM problem [8].”
BackgroundGNSS Odometry: Precise Trajectory Estimation Based on Carrier Phase Cycle Slip Estimation
Suzuki, T. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3182795 · Cited by 31
A unified framework for data association aware robust belief space planning and perception
Pathak, S., Thomas, A., & Indelman, V. (2018) · The International Journal of Robotics Research · doi:10.1177/0278364918759606 · Cited by 31
“…years, the SLAM research community has investigated approaches to be resilient to false data association (outliers) overlooked by front-end algorithms (e.g. image matching); see, e.g. Carlone et al. (2014), Indelman et al. (2014b, 2016), Olson and Agarwal (2013), and Sunderhauf and Protzel (2012).”
BackgroundVisual Place Recognition for Autonomous Mobile Robots
Horst, M., & Möller, R. (2017) · Robotics · doi:10.3390/robotics6020009 · Cited by 25
Robust Estimation and Applications in Robotics
Bosse, M. C., Agamennoni, G., & Gilitschenski, I. (2016) · now publishers, Inc. eBooks · doi:10.1561/9781680832150 · Cited by 25
Sun, L., & Deng, L. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3152837 · Cited by 28
“[24]–[26]) requires the initial guess, so they could easily converge to local minima if the initialization is not good enough.”
BackgroundRecent advances on simultaneous localization and mapping for mobile robots
Zamora, E., & Yu, W. (2013) · IETE Technical Review · doi:10.4103/0256-4602.125671 · Cited by 23
“In order to prevent the diver ‑ gence of optimization process due to data association errors, Sünderhauf and Protzel [50] reformulated the objective function, allowing the optimization process to change parts of the topological structure of the graph.”
MethodResolving Marker Pose Ambiguity by Robust Rotation Averaging with Clique Constraints
Chng, S.-F., Sogi, N., Purkait, P., Chin, T.-J., & Fukui, K. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196902 · Cited by 22
“To enable an efficient algorithm for (18), we apply the lifting approach [24].”
MethodLoop Closing Detection in RGB-D SLAM Combining Appearance and Geometric Constraints
Zhang, H., Liu, Y., & Tan, J. (2015) · Sensors · doi:10.3390/s150614639 · Cited by 21
“Requirement for precision is very high [6], which is close to 100%.”
BackgroundInter-Robot Range Measurements in Pose Graph Optimization
Boroson, E. R., Hewitt, R. A., Ayanian, N., & de la Croix, J.-P. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341227 · Cited by 21
“Sünderhauf and Protzel originally proposed this formulation [26], as it is nearly a step function between 0 and 1, but is differentiable everywhere.”
BackgroundMethodInfluentialA Multi-Hypothesis Approach to Pose Ambiguity in Object-Based SLAM
Fu, J., Huang, Q., Doherty, K., Wang, Y., & Leonard, J. J. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9635956 · Cited by 24
“S¨underhauf et al. [19], [20] develop methods to change the graph topology and achieve improved robustness to out-liers with switchable constraints.”
BackgroundTowards a Fully Automated 3D Reconstruction System Based on LiDAR and GNSS in Challenging Scenarios
Ren, R., Fu, H., Xue, H., Sun, Z., Ding, K., & Wang, P. (2021) · Remote Sensing · doi:10.3390/rs13101981 · Cited by 20
“To eliminate the influence of these outliers, robust optimization algorithms have been designed, such as the Switchable Constraints [28], the Max-Mixture Model [29], etc.”
MethodEstimation Contracts for Outlier-Robust Geometric Perception
Carlone, L. (2023) · Foundations and Trends in Robotics · doi:10.1561/2300000077 · Cited by 34
Agarwal, P., Burgard, W., & Stachniss, C. (2014) · IEEE Robotics & Automation Magazine · doi:10.1109/mra.2014.2322282 · Cited by 25
“DCS can be formulated as a generalization of switchable constraints [12], another state-of-the-art back end for robust operation in the presence of outliers.”
BackgroundLowry, S. M., Wyeth, G. F., & Milford, M. J. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6906932 · Cited by 18
“However, methods such as [30, 31] allow correction of occasional false positive loop closures using topological information.”
MethodGraph SLAM Built over Point Clouds Matching for Robot Localization in Tunnels
Sesmero, C. P., Lorente, S. V., & Castro, M. D. (2021) · Italian National Conference on Sensors · doi:10.3390/s21165340 · Cited by 17
“In order to solve the problem of graph-based approaches, the authors in [6–8] presented non-linear optimization on constraint graphs.”
Aasim, W. F. A. W., Okasha, M., & Faris, W. F. (2022) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-022-01643-y · Cited by 18
“The edge between these points can be described as the full transformation from xi to xj [67, 68].”
BackgroundAn Autonomous Forklift with 3D Time-of-Flight Camera-Based Localization and Navigation
Behrje, U., Himstedt, M., & Maehle, E. (2018) · International Conference on Control, Automation, Robotics and Vision (ICARCV) · doi:10.1109/icarcv.2018.8581085 · Cited by 17
“This is accompanied by a robust outlier detection based on switchable constraints [25].”
BackgroundA review of optimisation strategies used in simultaneous localisation and mapping
Huang, S. (2018) · Journal of Control and Decision · doi:10.1080/23307706.2018.1552207 · Cited by 16
“The first interesting work along this line is by Sunderhauf & Protzel (2012), where additional variables are added to indicate whether a measurement is outlier or not.”
BackgroundGlobally Optimal Fetoscopic Mosaicking Based on Pose Graph Optimisation With Affine Constraints
Li, L., Bano, S., Deprest, J. A., David, A. L., Stoyanov, D. V., & Vasconcelos, F. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3100938 · Cited by 16
“Robustness to outliers of loop closures was investigated in [23], [24].”
BackgroundEfficient 2D Graph SLAM for Sparse Sensing
Zhou, H., Hu, Z., Liu, S., & Khan, S. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981200 · Cited by 21
“4) Graph optimization (line 4-5): The pose graph ob-jective follows the classical formulation as introduced by S¨underhauf et al. [34], which is given by where e o is identical to (10), and where w ij represents a rigid transformation between pose y i and y j calculated by the loop closure detector.”
MethodGPU accelerated graph SLAM and occupancy voxel based ICP for encoder-free mobile robots
Ratter, A., Sammut, C., & McGill, M. (2013) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2013.6696404 · Cited by 17
“More complex optimisation techniques, such as by Sunderhauf and Protzel [14], have recently been described and could easily be used as part of our SLAM system, however our results show that the added computational expense is not necessary in many real world environments.”
MethodIntegrity-Constrained Factor Graph Optimization for GNSS Positioning in Urban Canyons
Xia, X., Wen, W., & Hsu, L. (2024) · NAVIGATION Journal of the Institute of Navigation · doi:10.33012/navi.660 · Cited by 15
CrowdFusion: Multisignal Fusion SLAM Positioning Leveraging Visible Light
Li, Z., Zhao, X., Zhao, Z., & Braun, T. I. (2023) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3260205 · Cited by 15
“Therefore, the conditional probability over all variables and constraints is expressed by [36] (11) where P ( x i + 1 | x i , s i ) and P ( x j | x i , s ij ) are called factors to encode probabilistic constraints over the corresponding nodes.”
BackgroundA brief review of simultaneous localization and mapping
Kong, Z., & Lu, Q. (2017) · Annual Conference of the IEEE Industrial Electronics Society (IECON) · doi:10.1109/iecon.2017.8216955 · Cited by 15
“Graph optimization method [7] consists of graph construction and graph optimization.”
MethodGraphTinker: Outlier rejection and inlier injection for pose graph SLAM
Xie, L., Wang, S., Markham, A., & Trigoni, N. (2017) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2017.8206596 · Cited by 14
“Sünderhauf and Protzel [1], [2] indicate that the topology of a factor graph can be partially unfixed and present the idea of Switchable Constraints (SC) where a switchable variable is given to each loop closure constraint.”
BackgroundMethodInfluentialClosed-form Online Pose-chain SLAM
Dubbelman, G., & Brett Browning, A. (2013) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2013.6631319 · Cited by 14
“…is to use filter based solutions, e.g. EKF-SLAM, UKF-SLAM or FAST-SLAM [23], when computational resources are limited and to use more accurate graph based [9], [10], [12], [15], [19], [22], [24] or bundle adjustment based [11], [13] solutions when sufficient computational resources are available.”
MethodA Variational Feature Encoding Method of 3D Object for Probabilistic Semantic SLAM
Yu, H., & Lee, B. H. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593831 · Cited by 15
Submap-Based Pose-Graph Visual SLAM: A Robust Visual Exploration and Localization System
Chen, W., Zhu, L., Guan, Y., Kube, C. R., & Zhang, H. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8594097 · Cited by 16
AEROS: AdaptivE RObust Least-Squares for Graph-Based SLAM
Ramezani, M., Mattamala, M., & Fallon, M. F. (2022) · Frontiers in Robotics and AI · doi:10.3389/frobt.2022.789444 · Cited by 15
“An efficient approach when dealing with false-positive loop-closures in the back-end is the approach proposed by 4 Sünderhauf and Protzel [11] in which, for each loop-closure added to the factor graph, a switch variable is added to reweight the cost function: {X ,S}MAP = argmin X ,S ∑ i ‖fi(Xi,Xi−1)− yi‖2Σi + ∑ j ( w…”
BackgroundMethodResultInfluentialHu, E., & Sun, L. (2022) · Information Sciences · doi:10.1016/j.ins.2022.10.086 · Cited by 14
Practical, Fast and Robust Point Cloud Registration for Scene Stitching and Object Localization
Sun, L. (2022) · IEEE Access · doi:10.1109/access.2021.3140105 · Cited by 12
“Iterative local solvers [26]–[28] have been applied to optimize M-estimation problems in earlier time, but they require initial guess and may get trapped into local minima.”
Robust linear pose graph-based SLAM
Cheng, J., Kim, J., Shao, J., & Zhang, W. (2015) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2015.04.010 · Cited by 13
Low-cost Lane-level Positioning in Urban Area Using Optimized Long Time Series GNSS and IMU Data
MEGURO, J., Arakawa, T., Mizutani, S., & Takanose, A. (2018) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2018.8569565 · Cited by 13
“However, sometimes SLAM/map localization using Lidar fails [7].”
BackgroundFree-Resolution Probability Distributions Map-Based Precise Vehicle Localization in Urban Areas
Kim, K., & Jee, G.-I. (2020) · Sensors · doi:10.3390/s20041220 · Cited by 14
LIV-LAM: LiDAR and Visual Localization and Mapping
Radmanesh, R., Wang, Z., Chipade, V. S., Tsechpenakis, G., & Panagou, D. (2020) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc45564.2020.9148037 · Cited by 12
“The front-end can detect these loop closures based on visual recognition (sensory vision), or by LiDAR as in [20], [3], [11].”
MethodGraph-Based Cooperative Localization Using Symmetric Measurement Equations
Gulati, D., Zhang, F., Clarke, D., & Knoll, A. (2017) · Sensors · doi:10.3390/s17061422 · Cited by 12
“Incorrect values in the graph degrade the optimization [35].”
BackgroundOutlier Accommodation in Nonlinear State Estimation: A Risk-Averse Performance- Specified Approach
Aghapour, E., Rahman, F., & Farrell, J. A. (2020) · IEEE Transactions on Control Systems Technology · doi:10.1109/tcst.2019.2952529 · Cited by 13
“Recently, new optimization-based outlier accommodation approaches have been suggested [5], [11]–[13].”
BackgroundMethodCeiling vision-based active SLAM framework for dynamic and wide-open environments
An, S.-Y., Lee, L.-K., & Oh, S. (2015) · Autonomous Robots · doi:10.1007/s10514-015-9453-0 · Cited by 12
“Thus our system is similar to a pose graph (Sunderhauf and Protzel 2012) but state update is done by standard IEKF framework (Estrada et al.”
MethodRobust simultaneous localization and mapping via information matrix estimation
Graham, M. C., & How, J. P. (2014) · IEEE/ION Position Location and Navigation Symposium · doi:10.1109/plans.2014.6851458 · Cited by 14
“weights for each of the loop closure measurements [5,8].”
BackgroundMethodInfluentialIntrospective Evaluation of Perception Performance for Parameter Tuning without Ground Truth
Hu, H., & Kantor, G. A. (2017) · Robotics: Science and Systems · doi:10.15607/rss.2017.xiii.033 · Cited by 10
Metadata
- Authors
- Sunderhauf, N., & Protzel, P.
- Affiliations
- Chemnitz University of Technology (DE)
- ORCID
- Protzel, Peter
- Venue
- 2012 IEEE International Conference on Robotics and Automation
- Publisher
- IEEE
- Published
- 2012-05
- Pages
- 1254-1261
- ISSN
- 1050-4729 (ISSN-L)
- OpenAlex
- W1991172434
- Cited by
- 198
- References
- 33
- Citations by year
- 2026: 4 · 2025: 10 · 2024: 10 · 2023: 14 · 2022: 19 · 2021: 12 · 2020: 19 · 2019: 13 · 2018: 23 · 2017: 10 · 2016: 12 · 2015: 13 · 2014: 16 · 2013: 13 · 2012: 10
- FWCI
- 30.98
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Proceedings - IEEE International Conference on Robotics and Automation/Proceedings (publishedVersion)
- Funders
- Technische Universität Chemnitz
- Queensland University of Technology
- University of Queensland
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Robotics and Sensor-Based Localization
- Physical Sciences › Computer Science › Artificial Intelligence › Target Tracking and Data Fusion in Sensor Networks
- Physical Sciences › Earth and Planetary Sciences › Geology › 3D Surveying and Cultural Heritage
- Keywords
- pose graph SLAM, back-end optimization, loop closure detection, robust optimization, sparse optimization, topological correctness
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.