SeqSLAM: Visual route-based navigation for sunny summer days and stormy winter nights
The DOI matches its Crossref record; 20 of the 22 references it cites were found in a registry or an index.
Abstract
Learning and then recognizing a route, whether travelled during the day or at night, in clear or inclement weather, and in summer or winter is a challenging task for state of the art algorithms in computer vision and robotics. In this paper, we present a new approach to visual navigation under changing conditions dubbed SeqSLAM. Instead of calculating the single location most likely given a current image, our approach calculates the best candidate matching location within every local navigation sequence. Localization is then achieved by recognizing coherent sequences of these “local best matches”. This approach removes the need for global matching performance by the vision front-end - instead it must only pick the best match within any short sequence of images. The approach is applicable over environment changes that render traditional feature-based techniques ineffective. Using two car-mounted camera datasets we demonstrate the effectiveness of the algorithm and compare it to one of the most successful feature-based SLAM algorithms, FAB-MAP. The perceptual change in the datasets is extreme; repeated traverses through environments during the day and then in the middle of the night, at times separated by months or years and in opposite seasons, and in clear weather and extremely heavy rain. While the feature-based method fails, the sequence-based algorithm is able to match trajectory segments at 100% precision with recall rates of up to 60%.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“This has led to develop new methods that explicitly account for such variations by matching sequences [173], gathering different visual appearances into a unified representation [48], or using spatial as well as appearance information [106].”
From Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age (2016) · cites this paper as [173]
Citations
Cites
Object recognition from local scale-invariant features
Lowe, D. G. (1999) · Proceedings of the Seventh IEEE International Conference on Computer Vision · doi:10.1109/iccv.1999.790410 · Cited by 16,472 · cited 3×
“Many place recognition techniques rely on feature-finding algorithms such as SIFT [9] and SURF [10] which, despite their impressive rotation and scale invariant properties, are inherently unsuitable when dealing with extreme perceptual M.J. Milford and G.F. Wyeth are with the School of Electrical Engineering and Compu…”
MethodSURF: Speeded Up Robust Features
Bay, H., Tuytelaars, T., & Van Gool, L. (2006) · Lecture Notes in Computer Science · doi:10.1007/11744023_32 · Cited by 14,955
MonoSLAM: 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×
“isual recognition of places and place sequences in the field of robotic navigation has advanced rapidly in recent years, with several techniques [1-5] now being competitive with laser range-based techniques.”
BackgroundA solution to the simultaneous localization and map building (SLAM) problem
Dissanayake, M. W. M. G., Newman, P., Clark, S., Durrant-Whyte, H. F., & Csorba, M. (2001) · IEEE Transactions on Robotics and Automation · doi:10.1109/70.938381 · Cited by 2,661
Matching Local Self-Similarities across Images and Videos
Shechtman, E., & Irani, M. (2007) · 2007 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2007.383198 · Cited by 1,076 · cited 1×
“One promising step [15] towards reducing the reliance on consistent feature detection uses the self-similarity of objects within a frame or video sequence to robustly recognize objects under varying conditions.”
BackgroundIncremental mapping of large cyclic environments
Gutmann, J.-S., & Konolige, K. (2003) · Proceedings 1999 IEEE International Symposium on Computational Intelligence in Robotics and Automation. CIRA'99 (Cat. No.99EX375) · doi:10.1109/cira.1999.810068 · Cited by 568 · cited 1×
“There have been attempts to increase the robustness of visual techniques to environment change by performing sensor fusion with lasers at the level of landmarks [13] or dense depth maps [14].”
BackgroundFrameSLAM: From Bundle Adjustment to Real-Time Visual Mapping
Konolige, K., & Agrawal, M. (2008) · IEEE Transactions on Robotics · doi:10.1109/tro.2008.2004832 · Cited by 532
RatSLAM: a hippocampal model for simultaneous localization and mapping
Milford, M. J., Wyeth, G. F., & Prasser, D. (2004) · IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA '04. 2004 · doi:10.1109/robot.2004.1307183 · Cited by 399 · cited 6×
“The approach taken in this work is partly inspired by the author's previous work in vision-based mapping in RatSLAM [16] [17] [18].”
MethodInfluentialSelf-citationMapping a Suburb With a Single Camera Using a Biologically Inspired SLAM System
Milford, M. J., & Wyeth, G. F. (2008) · IEEE Transactions on Robotics · doi:10.1109/tro.2008.2004520 · Cited by 378 · cited 7×
“The approach taken in this work is partly inspired by the author's previous work in vision-based mapping in RatSLAM [16] [17] [18].”
BackgroundMethodInfluentialSelf-citationDynamic Maps for Long-Term Operation of Mobile Service Robots
Biber, P., & Duckett, T. (2005) · Robotics: Science and Systems I · doi:10.15607/rss.2005.i.003 · Cited by 334 · cited 1×
“The problem of visual place recognition across time has received far less attention, with relatively few attempts to address the problem [6, 7], including high dynamic range approaches [8].”
Background
Cited by
Items 201–300 of 987
Wang, K., Guo, J. H., Chen, K., & Lǚ, J. (2025) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2025.3545479 · Cited by 42
Addressing Challenging Place Recognition Tasks Using Generative Adversarial Networks
Latif, Y., Garg, R., Milford, M. J., & Reid, I. D. (2018) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2018.8461081 · Cited by 36
Self-citationGarg, S., Babu, M., V, Dharmasiri, T., Hausler, S., Sünderhauf, N., Kumar, S., Drummond, T. W., & Milford, M. J. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8794178 · Cited by 36
“In such cases, the above finding often gets discounted because the overall structure of the scene remains mostly similar that massively aids in matching, especially when images are down-sampled, for example, patch-normalization in SeqSLAM [6], HoG in [7], and flattened deep-learnt tensors in [51], [52], [46], [36].”
BackgroundMethodInfluentialSelf-citationSemantic loop closure detection based on graph matching in multi-objects scenes
Qin, C., Zhang, Y., Liu, Y., & Lv, G. (2021) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2021.103072 · Cited by 42
Accurate localization by fusing images and GPS signals
Kumar, V., Jawahar, C. V., & Chari, V. (2015) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2015.7301390 · Cited by 35
Expanding the Limits of Vision‐based Localization for Long‐term Route‐following Autonomy
Paton, M., Pomerleau, F., MacTavish, K., Ostafew, C. J., & Barfoot, T. D. (2016) · Journal of Field Robotics · doi:10.1002/rob.21669 · Cited by 38
“First introduced by Milford and Wyeth (2012) as SeqSLAM, this method first computes a confusion matrix between an array of recent images and the array of map images using whole image matching, and then it searches for the best diagonal path through the matrix. While effective, this method is highly susceptible to scal…”
MethodInfluentialSelf-Driving Vehicles: Key Technical Challenges and Progress Off the Road
Milford, M. J., Anthony, S., & Scheirer, W. J. (2019) · IEEE Potentials · doi:10.1109/mpot.2019.2939376 · Cited by 35
Self-citationMulti-Session Visual SLAM for Illumination-Invariant Re-Localization in Indoor Environments
Labbé, M., & Michaud, F. (2022) · Frontiers in Robotics and AI · doi:10.3389/frobt.2022.801886 · Cited by 40
“To avoid using multiple experiences of the same locations, SeqSLAM [3], [4] matches sequences of visual frames instead of trying to localize robustly each individual frame against the map.”
MethodVision‐based Simultaneous Localization and Mapping in Changing Outdoor Environments
Milford, M. J., Vig, E., Scheirer, W. J., & Cox, D. (2014) · Journal of Field Robotics · doi:10.1002/rob.21532 · Cited by 34
“The previous state-of-the-art approach SeqSLAM solved this challenge by matching over long sequences of images (Milford & Wyeth, 2012), rather than single images.”
BackgroundMethodInfluentialSelf-citationOpenSeqSLAM2.0: An Open Source Toolbox for Visual Place Recognition Under Changing Conditions
Talbot, B., Garg, S., & Milford, M. J. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593761 · Cited by 34
Self-citationCross-Weather Image Alignment via Latent Generative Model With Intensity Consistency
Zhou, H., Ma, J., Tan, C. C., Zhang, Y., & Ling, H. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2980210 · Cited by 37
“Robotics and intelligent transportation systems research community have worked on similar problems related to cross-weather road scenes, such as robust scene descriptors (BoW [20], GIST [21], and CNN [22]).”
BackgroundMethodSemantic Segmentation With Low Light Images by Modified CycleGAN-Based Image Enhancement
Cho, S. W., Baek, N. R., Koo, J. H., Arsalan, M., & Park, K. R. (2020) · IEEE Access · doi:10.1109/access.2020.2994969 · Cited by 36
“The road scene open databases captured in the existing nighttime environment include Berkeley deep drive 100K (BDD100K) [34], Nighttime Driving [12], and Alderley [35] databases.”
BackgroundKeroglou, C., Kansizoglou, I., Michailidis, P., Oikonomou, K. M., Papapetros, I. T., Dragkola, P., Michailidis, I., Gasteratos, A. C., Kosmatopoulos, E. B., & Sirakoulis, G. C. (2023) · IEEE Transactions on Medical Robotics and Bionics · doi:10.1109/tmrb.2023.3261342 · Cited by 36
“On the other hand dynamic conditions such as different illumination states, weather conditions or moved objects between the mapping and the query observations can significantly affect the overall performance [57], [58], [59], [60], [61].”
BackgroundMUI-TARE: Cooperative Multi-Agent Exploration With Unknown Initial Position
Yan, J., Lin, X., Ren, Z., Zhao, S., Yu, J., Cao, C., Yin, P., Zhang, J., & Scherer, S. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3281262 · Cited by 34
“To merge sub-maps generated by each agent accurately and robustly, the SLAM community has developed many methods by leveraging distinguishable semantic objects [11], [26], bag-of-words vectors [27], learned full-image descriptors [28], and sequence-based place recognition [12], [29], to name a few.”
MethodTowards improving the efficiency of sequence-based SLAM
Liu, Y., & Zhang, H. (2013) · IEEE International Conference on Mechatronics and Automation (ICMA) · doi:10.1109/icma.2013.6618095 · Cited by 34
“A reasonable and intuitive way as in [7] is to correspond each Z t ,j to its closest counterpart in Xt and this can be done via interpolation after the two sequences are aligned with each other.”
BackgroundMethodInfluentialGlobal image signature for visual loop-closure detection
Carrasco, P. L. N., Font, F. B., & Oliver, G. (2015) · Autonomous Robots · doi:10.1007/s10514-015-9522-4 · Cited by 39
“Appearance-based the signature is calculated from texture, color information, transformations in the frequency space or matrix factorizations (Venkatesan et al. 2000; Lin et al. 2001; Sunderhauf and Protzel 2011; Milford and Wyeth 2012).”
MethodRobust visual localization in changing lighting conditions
Kim, P., Coltin, B., Alexandrov, O., & Kim, H. J. (2017) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2017.7989640 · Cited by 37
Robust Place Recognition and Loop Closing in Laser-Based SLAM for UGVs in Urban Environments
Cao, F., Zhuang, Y., Zhang, H., & Wang, W. (2018) · IEEE Sensors Journal · doi:10.1109/jsen.2018.2815956 · Cited by 34
“SeqSLAM [26] is visual route-based navigation system, which calculates the best candidate matching location within every local navigation sequence instead of calculating the single location most likely given a current image.”
BackgroundSchubert, S., Neubert, P., & Protzel, P. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197044 · Cited by 33
“Since all methods take as input descriptors and return more condition-invariant descriptors, they can be easily combined with available sequence-based methods (e.g., SeqSLAM [31], HMM [32], ABLE-M [33], MCN [34]) or experience maps [15].”
BackgroundRadar-to-Lidar: Heterogeneous Place Recognition via Joint Learning
Yin, H., Xu, X., Wang, Y., & Xiong, R. (2021) · Frontiers in Robotics and AI · doi:10.3389/frobt.2021.661199 · Cited by 39
Season-Invariant and Viewpoint-Tolerant LiDAR Place Recognition in GPS-Denied Environments
Cao, F., Yan, F., Wang, S., Zhuang, Y., & Wang, W. (2020) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2019.2962416 · Cited by 33
“11, our method achieves better precision–recall performances compared with SeqSLAM under these challenging perceptual conditions.”
MethodInfluentialLocation graphs for visual place recognition
Stumm, E. S., Christopher Mei, A., Lacroix, S., & Chli, M. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139964 · Cited by 33
“For completeness, we also include results from the commonly used SeqSLAM framework [21], although the approach differs drastically from the one presented here.”
MethodA Multi-Domain Feature Learning Method for Visual Place Recognition
Yin, P., Xu, L., Li, X.-Q., Chen, Y., Li, Y., Srivatsan, R. A., Li, L., Ji, J., & He, Y. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8793752 · Cited by 36
“Eq.”
BackgroundMethodInfluentialA Survey on Monocular Re-Localization: From the Perspective of Scene Map Representation
Miao, J., Jiang, K., Wen, T., Wang, Y., Jia, P., Wijaya, B., Zhao, X., Qian, C., Xiao, Z., Huang, J. H., et al. (2024) · IEEE Transactions on Intelligent Vehicles · doi:10.1109/tiv.2024.3378716 · Cited by 55
“In the MRL solution, the data in the map is captured by sensors during a historical period, and it probably exhibits extreme variations in visual conditions (like weather [12], illumination [13], day-night transitions [14], etc.) and even modality [15], [16], compared to the currently observed data.”
InfluentialLearning place-dependant features for long-term vision-based localisation
McManus, C., Upcroft, B., & Newman, P. M. (2015) · Autonomous Robots · doi:10.1007/s10514-015-9463-y · Cited by 32
“These unique visual elements, called scene signatures , are used for realtime localisation to provide a rough metric estimate of the vehicle’s position and orientation (Milford 2013; Milford and Wyeth 2012), but is limited to topological localisation.”
BackgroundPepperell, E., Corke, P., & Milford, M. J. (2016) · The International Journal of Robotics Research · doi:10.1177/0278364915618766 · Cited by 32
“Existing condition-invariant algorithms typically employ global image descriptors with temporal information in the form of image sequences Milford and Wyeth, 2012; Naseer et al., 2014; Pepperell et al., 2014a; Sünderhauf et al., 2013b.”
BackgroundMethodResultSelf-citationMemorable Maps: A Framework for Re-Defining Places in Visual Place Recognition
Zaffar, M., Ehsan, S., Milford, M. J., & McDonald-Maier, K. D. (2020) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2020.3001228 · Cited by 40
Self-citationBenchmark of Visual SLAM Algorithms: ORB-SLAM2 vs RTAB-Map
Ragot, N., Khemmar, R., Pokala, A., Rossi, R., & Ertaud, J.-Y. (2019) · International Conference on Emerging Security Technologies · doi:10.1109/est.2019.8806213 · Cited by 37
“The most representative approach in this category is SeqSLAM [24] which has demonstrated a much better performance than FabMAP under conditions of extreme atmospheric changes and brightness.”
Chen, H., ZHANG, G., & Ye, Y. (2020) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2020.3010580 · Cited by 32
“Different from the methods mentioned above, SeqSLAM [6] group consecutive place images to recognize coherent sequences, which makes them tolerant to changes of weather and season.”
BackgroundMethodInfluentialHu, H., Wang, H., Liu, Z., & Chen, W. (2021) · IEEE/CAA Journal of Automatica Sinica · doi:10.1109/jas.2021.1003907 · Cited by 34
“addressed most cases of loop closure detection in real-time visual SLAM [29], [30] without significant environmental changes.”
BackgroundVisual Place Recognition using LiDAR Intensity Information
Giammarino, L. D., Aloise, I., Stachniss, C., & Grisetti, G. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636649 · Cited by 39
“SeqSLAM [28] aims at matching image sequences under seasonal changes and computes a matching matrix that stores the similarity between the images in a query sequence and a database.”
BackgroundEnhancing RGB-D SLAM Performances Considering Sensor Specifications for Indoor Localization
Bouazzaoui, I. E., Rodríguez, S., & Elouardi, A. (2021) · IEEE Sensors Journal · doi:10.1109/jsen.2021.3073676 · Cited by 32
Kong, D., Fang, Z., Hou, K., Li, H., Jiang, J., Coleman, S., & Kerr, D. (2022) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2022.3168892 · Cited by 32
“Among them, Milford et al. first tried to migrate SeqSLAM [19] to the event camera in 2018 and completed a relatively basic recognition experiment based on event frames [25].”
BackgroundMethodNeuromorphic sequence learning with an event camera on routes through vegetation
Zhu, L., Mangan, M., & Webb, B. H. (2023) · Science Robotics · doi:10.1126/scirobotics.adg3679 · Cited by 34
“Benchmarking against SeqSLAM and perfect memory SeqSLAM (35, 81) is a well-known VPR algorithm that calculates a locally enhanced difference matrix and selects the best matches only in a short sequence of images.”
MethodVisual Object Recognition with 3D-Aware Features in KITTI Urban Scenes
Yebes, J. J., Bergasa, L. M., & García-Garrido, M. Á. (2015) · Sensors · doi:10.3390/s150409228 · Cited by 36
“Therefore, to provide image understanding capabilities to autonomous vehicles, some of the urban challenges may include: object detection under occlusion [11], estimation of object orientation on 3D scenes [12], detection at far distances [13], determining the geometric layout of the scene [14], dealing with varying i…”
BackgroundSeqSphereVLAD: Sequence Matching Enhanced Orientation-invariant Place Recognition
Yin, P., Wang, F., Egorov, A., Hou, J., Zhang, J., & Choset, H. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341727 · Cited by 35
“Instead of relying on the single scan observation, SeqSLAM [5] utilizes a sequence of observations to improve the matching robustness, which significantly improves the place recognition accuracy under variant measurement noise and environmental conditions.”
MethodInfluentialVector Semantic Representations as Descriptors for Visual Place Recognition
Neubert, P., Schubert, S., Schlegel, K., & Protzel, P. (2021) · Robotics: Science and Systems · doi:10.15607/rss.2021.xvii.083 · Cited by 37
From dusk till dawn: Localisation at night using artificial light sources
Nelson, P., Churchill, W., Posner, I., & Newman, P. M. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139930 · Cited by 31
“One of the few examples of work that addresses the problem of navigating a vehicle at night using consumer cameras is SeqSLAM [12], in which localisation is achieved by recognising coherent sequences of heavily down-sampled images.”
BackgroundLoop Closure Detection Based on Multi-Scale Deep Feature Fusion
Chen, B., Yuan, D., Liu, C., & Wu, Q. H. (2019) · Applied Sciences · doi:10.3390/app9061120 · Cited by 31
Paton, M., MacTavish, K., Ostafew, C. J., & Barfoot, T. D. (2015) · 2015 IEEE International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra.2015.7139391 · Cited by 30
“Recent publications have provided the means for computing color-constant greyscale images from a three-channel camera [5], [14], and have been shown to be successful in localization and mapping systems [3], [8].”
Jacobson, A., Zeng, F., Smith, D., Boswell, N., Peynot, T., & Milford, M. J. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593750 · Cited by 30
“Visual place recognition systems for surface applications such as FABMAP [2] and SeqSLAM [4] or state-of-theart deep learning approaches [5] have demonstrated the capability of performing place recognition despite changing environmental and seasonal conditions [4], however these systems require a manual grounding in t…”
MethodInfluentialSelf-citationFerrarini, B., Milford, M. J., McDonald-Maier, K. D., & Ehsan, S. (2022) · IEEE Transactions on Robotics · doi:10.1109/tro.2022.3148908 · Cited by 42
“[6] used those layers’ features to improve the matching performance of SeqSLAM [36] under viewpoint changes.”
Self-citationSTA-VPR: Spatio-Temporal Alignment for Visual Place Recognition
Lu, F. H., Chen, B., Zhou, X., & Song, D. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3067623 · Cited by 30
Environment selection and hierarchical place recognition
Mohan, M., Gálvez‐López, D., Monteleoni, C., & Sibley, G. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139966 · Cited by 30
“Consequently, a large number of techniques for describing and comparing images and for retrieving the best matches have been proposed in the literature [3], [15].”
BackgroundRelative geometry-aware siamese neural network for 6DOF camera relocalization
Li, Q., Zhu, J., Cao, R., Sun, K., Garibaldi, J. M., Li, Q., Liu, B., & Qiu, G. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.09.071 · Cited by 30
Highly Robust Visual Place Recognition Through Spatial Matching of CNN Features
Camara, L. G., Gaebert, C., & Přeučil, L. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196967 · Cited by 30
“However, in a mobile robotics environment there will almost certainly exist some time correlation between frames, a fact that has been exploited by some successful VPR techniques [29, 30, 31].”
BackgroundLocal region detector + CNN based landmarks for practical place recognition in changing environments
Neubert, P., & Protzel, P. (2015) · European Conference on Mobile Robots · doi:10.1109/ecmr.2015.7324051 · Cited by 31
“Since the predicted images are not suitable for local keypoint extraction due to smoothing and artefacts, the same issues of holistic image matching as for SeqSLAM holds for this approach.”
MethodInfluentialDetecting the Expectancy of a Place Using Nearby Context for Appearance-Based Mapping
Kazmi, S. M. A. M., & Mertsching, B. (2019) · IEEE Transactions on Robotics · doi:10.1109/tro.2019.2926475 · Cited by 29
“0 [2], SeqSLAM [3], iBoW-LCD [23], and our previous work [9].”
MethodPersistent Stereo Visual Localization on Cross-Modal Invariant Map
Ding, X., Wang, Y., Xiong, R., Li, D., Tang, L., Yin, H., & Zhao, L. (2019) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2019.2942760 · Cited by 29
“While other studies try to discover the invariant subsets or robust representations of the maps for better long-term performance [2], [21]–[24].”
BackgroundResultRoute-following ants respond to alterations of the view sequence
Schwarz, S., Mangan, M., Webb, B. H., & Wystrach, A. (2020) · Journal of Experimental Biology · doi:10.1242/jeb.218701 · Cited by 30
Tracking‐DOSeqSLAM: A dynamic sequence‐based visual place recognition paradigm
Tsintotas, K. A., Bampis, L., Gasteratos, A. C., & FIET (2021) · IET Computer Vision · doi:10.1049/cvi2.12041 · Cited by 29
Loop-Closure Detection Using Local Relative Orientation Matching
Ma, J., Ye, X., Zhou, H., Mei, X., & Fan, F. (2021) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2021.3074520 · Cited by 31
PSE-Match: A Viewpoint-Free Place Recognition Method With Parallel Semantic Embedding
Yin, P., Xu, L., Feng, Z., Egorov, A., & Li, B. (2021) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2021.3102429 · Cited by 30
Robust Onboard Localization in Changing Environments Exploiting Text Spotting
Zimmerman, N., Wiesmann, L., Guadagnino, T., Läbe, T., Behley, J., & Stachniss, C. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981049 · Cited by 37
An Online Sparsity-Cognizant Loop-Closure Algorithm for Visual Navigation
Latif, Y., Huang, G. P., Leonard, J. J., & Neira, J. (2014) · Robotics: Science and Systems · doi:10.15607/rss.2014.x.036 · Cited by 29
“In particular, Milford and Wyeth [23] proposed SeqSLAM, which is able to localize with drastic lighting and weather changes by matching sequences of images with each other as opposed to single images.”
BackgroundMaddern, W., Stewart, A. D., & Newman, P. M. (2014) · IEEE Intelligent Vehicles Symposium · doi:10.1109/ivs.2014.6856471 · Cited by 28
“The importance of image sequence information when localising across changes in appearance is demonstrated by a number of approaches [18], [19].”
BackgroundMethodEvaluation of Object Proposals and ConvNet Features for Landmark-based Visual Place Recognition
Hou, Y., Zhang, H., & Zhou, S. (2017) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-017-0735-y · Cited by 28
“In general, local feature-based techniques [3] and global feature-based techniques [4–7] are the two most popular categories of these techniques.”
MethodAre State-of-the-art Visual Place Recognition Techniques any Good for Aerial Robotics?
Zaffar, M., Khaliq, A., Ehsan, S., Milford, M. J., Alexis, K., & McDonald-Maier, K. D. (2019) · Open Access at Essex (University of Essex) · doi:10.48550/arxiv.1904.07967 · Cited by 31
Self-citationWang, Y., Qiu, Y., Cheng, P., & Duan, X. (2020) · Remote Sensing · doi:10.3390/rs12233890 · Cited by 29
“SeqSLAM [21] considers a short image sequence instead of a single image to solve perceptual aliasing.”
BackgroundSpiking Neural Networks for Visual Place Recognition Via Weighted Neuronal Assignments
Hussaini, S., Milford, M. J., & Fischer, T. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3149030 · Cited by 35
“This section provides a comparison of our proposed SNN model with the standard VPR methods, namely the simplistic Sum of Absolute Differences (SAD) [17] and a state-of-the-art approach, NetVLAD [16].”
InfluentialSelf-citationSectionKey: 3-D Semantic Point Cloud Descriptor for Place Recognition
Jin, S., Wu, Z., Zhao, C., Zhang, J., Peng, G., & Wang, D. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981605 · Cited by 28
“Since the sensor feedback is time-continuous in a SLAM system [25], a single loop closure occurrence often indicates high similarity on the adjacent LiDAR scans.”
iSimLoc: Visual Global Localization for Previously Unseen Environments With Simulated Images
Yin, P., Cisneros, I., Zhao, S., Zhang, J., Choset, H., & Scherer, S. (2023) · IEEE Transactions on Robotics · doi:10.1109/tro.2023.3238201 · Cited by 32
“[22] Place Recognition Hybrid Changing Known Low N/A No”
BackgroundPlace recognition based on matching of planar surfaces and line segments
Cupec, R., Nyarko, E. K., Filko, D., Kitanov, A., & Petrovic, I. K. (2015) · The International Journal of Robotics Research · doi:10.1177/0278364914548708 · Cited by 30
“…for application within a SLAM system that exploits prior probability by using sequence information like those presented by Kawewong et al. (2011), Milford and Wyeth (2012) and Milford (2013), this paper is focused on place recognition from a single image, that is, usage of any information about…”
MethodSelf-localization from images with small overlap
Kanji, T. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759662 · Cited by 31
“More recently, in [2], the authors demonstrated that DCNN features outperform the majority of the existing global features in typical VPR tasks.”
MethodResultInfluentialUnifying Visual Localization and Scene Recognition for People With Visual Impairment
Cheng, R., Wang, K., Bai, J. Y., & Xu, Z. (2020) · IEEE Access · doi:10.1109/access.2020.2984718 · Cited by 28
“score of the latent pair is evaluated by windowed uniqueness thresholding [37] to remove low-confidencematching results.”
MethodVisual Appearance Analysis of Forest Scenes for Monocular SLAM
Garforth, J., & Webb, B. H. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8793771 · Cited by 35
“This paper looks at the problem of performing visual Simultaneous Localisation and Mapping (SLAM) in unstructured natural environments (such as forests), rather than the structured man-made environments (such as offices and city streets) that play host to the majority of SLAM research [1], [2], [3].”
BackgroundSemSegMap – 3D Segment-based Semantic Localization
Cramariuc, A., Tschopp, F., Alatur, N., Benz, S., Falck, T., Brühlmeier, M., Hahn, B. P., Nieto, J. I., & Siegwart, R. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636156 · Cited by 40
“However, outdoor applications typically pose challenges to those methods, namely large-scale environments, self similarity and vast appearance changes due to weather, daytime and seasonal conditions [4].”
BackgroundSVG-Loop: Semantic–Visual–Geometric Information-Based Loop Closure Detection
Yuan, Z., Xu, K., Zhou, X., Deng, B., & Ma, Y. (2021) · Remote Sensing · doi:10.3390/rs13173520 · Cited by 27
“Tsintotas [38] proposed a sequence-based loop closure detection approach fusing the SeqSLAM and Bag-of -Words methods.”
InfluentialImproving condition- and environment-invariant place recognition with semantic place categorization
Garg, S., Jacobson, A., Kumar, S., & Milford, M. J. (2017) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2017.8206608 · Cited by 33
Self-citationVIZARD: Reliable Visual Localization for Autonomous Vehicles in Urban Outdoor Environments
Bürki, M., Schaupp, L., Dymczyk, M., Dubé, R., Cadena, C., Siegwart, R., & Nieto, J. I. (2019) · IEEE Intelligent Vehicles Symposium · doi:10.1109/ivs.2019.8814017 · Cited by 30
“nt accuracy to prevent deviation from the road lane boundaries. Note that there has also been a substantial amount of work on global localization in the realm of place recognition, or image retrieval [13]–[17]. These approaches, however, only provide a best matching image candidate in a map, instead of a 6DoF metric p…”
BackgroundScene structure registration for localization and mapping
Fernández-Moral, E., Rives, P., Arévalo, V., & González-Jiménez, J. (2015) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2015.09.009 · Cited by 26
“In this sense, our work is more related to visual place recognition approaches which exploit the information of a sequence of images to boost the recognition accuracy [13,14].”
BackgroundImage-Based Localization Aided Indoor Pedestrian Trajectory Estimation Using Smartphones
Zhou, Y., Zheng, X., Chen, R., Xiong, H., & Guo, S. (2018) · Sensors · doi:10.3390/s18010258 · Cited by 27
“When combined with the PDR trajectory, we can obtain a bias-free”
MethodSequence-based sparse optimization methods for long-term loop closure detection in visual SLAM
Han, F., Wang, H., Huang, G. P., & Zhang, H. (2018) · Autonomous Robots · doi:10.1007/s10514-018-9736-3 · Cited by 26
“Visual features of images are applied to represent different places in visual SLAM, including local features (Angeli et al. 2008; Cummins and Newman 2009; Labbe and Michaud 2014; Mur-Artal et al. 2015) and global features (Arroyo et al. 2015; Latif et al. 2014; Milford and Wyeth 2012).”
BackgroundMethodInfluentialRetrieval-based Localization Based on Domain-invariant Feature Learning under Changing Environments
Hu, H., Wang, H., Liu, Z., Yang, C., Chen, W., & Xie, L. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968047 · Cited by 30
“s real-time V-SLAM systems [15], but they fail if the query and database images are taken under tremendously different environments due to the mismatch of descriptors between them. DBoW2 [16])SeqSLAM [17] uses sequence information to avoid such failures but brings concerns about failure under a large variance of persp…”
MethodHierarchical visual localization for visually impaired people using multimodal images
Cheng, R., Hu, W., Chen, H., Fang, Y., Wang, K., Xu, Z., & Bai, J. Y. (2020) · Expert Systems with Applications · doi:10.1016/j.eswa.2020.113743 · Cited by 27
“This paper follows the parameter settings used in OpenSeqSLAM2.”
BackgroundMethodInfluentialDetection of loop closure in SLAM: A DeconvNet based approach
Mukherjee, A., Chakraborty, S., & Saha, S. K. (2019) · Applied Soft Computing · doi:10.1016/j.asoc.2019.04.041 · Cited by 25
Piasco, N., Sidibé, D., Gouet-Brunet, V., & Demonceaux, C. (2020) · International Journal of Computer Vision · doi:10.1007/s11263-020-01363-6 · Cited by 32
“VBL is exploited in various applications such as autonomous driving (Milford and Wyeth 2012), robot navigation or SLAM loop closing”
Voxel-Based Representation Learning for Place Recognition Based on 3D Point Clouds
Siva, S., Nahman, Z., & Zhang, H. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9340992 · Cited by 29
“It has been observed that, while performing long-term place recognition, global features outperform local features [7], [27], [28].”
BackgroundHigh-level visual features for underwater place recognition
Li, J., Eustice, R. M., & Johnson‐Roberson, M. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139706 · Cited by 31
Bio-inspired homogeneous multi-scale place recognition
Chen, Z., Lowry, S. M., Jacobson, A., Hasselmo, M. E., & Milford, M. J. (2015) · Neural Networks · doi:10.1016/j.neunet.2015.10.002 · Cited by 27
“7(b)) which is a large 70 km road-based dataset (2 × 35 km traverses) used in the FAB-MAP (Cummins & Newman, 2009) and SeqSLAM studies (Milford, 2013; Milford & Wyeth, 2012).”
Self-citationVisual Place Recognition for Autonomous Mobile Robots
Horst, M., & Möller, R. (2017) · Robotics · doi:10.3390/robotics6020009 · Cited by 25
“Holistic methods are also used in SeqSLAM [5,19], where the sum of absolute differences between low-resolution, patch-normalized images is computed.”
MethodGriffith, S., Chahine, G., & Pradalier, C. (2017) · The International Journal of Robotics Research · doi:10.1177/0278364917730606 · Cited by 25
“…an unstructured environment as it evolved week-to-week over three years (see e.g. Dong et al., 2017; Geiger et al., 2013; Glover et al., 2010; Milford and Wyeth, 2012; Milford et al., 2014; Naseer et al., 2017; Skrede, 2013; Sunderhauf et al., 2015; Valada et al., 2016, for examples of more…”
BackgroundAutomated Map Reading: Image Based Localisation in 2-D Maps Using Binary Semantic Descriptors
Panphattarasap, P., & Calway, A. D. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8594253 · Cited by 25
Don’t Worry About the Weather: Unsupervised Condition-Dependent Domain Adaptation
Porav, H., Bruls, T., & Newman, P. M. (2019) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2019.8917073 · Cited by 25
“Topological Localization: Similarly, widely used topological localization frameworks such as FABMAP [18], SeqSlam [19] or NetVLAD [20] achieve high recall and precision figures on clear, daytime images or when explicitly matching images with the same condition (e.g. winter-winter matching), but break down when the con…”
MethodPoint-Cloud-Based Place Recognition Using CNN Feature Extraction
Sun, T., Liu, M., Ye, H., & Yeung, D. (2019) · IEEE Sensors Journal · doi:10.1109/jsen.2019.2937740 · Cited by 25
“the assumption that the vehicle will repeat the same path rather than just revisit a single place, sequence matching is adopted [1], [3]–[7]; [4] propose a method to correct the wrong recognition when new information comes; the time consumed”
MethodBansal, A., Badino, H., & Huber, D. (2014) · IEEE Intelligent Vehicles Symposium · doi:10.1109/ivs.2014.6856605 · Cited by 33
“[17] developed a robust localization algorithm adding temporal filtering to an image template matching technique.”
BackgroundMethodCross-season place recognition using NBNN scene descriptor
Kanji, T. (2015) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2015.7353453 · Cited by 24
“On the other hand, robust place recognition has been achieved using global feature approaches [4], whereby each input scene is described by a single global feature vector.”
BackgroundA model for navigation in unknown environments based on a reservoir of hippocampal sequences
Leibold, C. (2020) · Neural Networks · doi:10.1016/j.neunet.2020.01.014 · Cited by 28
SeqMatchNet: Contrastive Learning with Sequence Matching for Place Recognition & Relocalization
Garg, S., Vankadari, M. B., & Milford, M. J. (2021) · Conference on Robot Learning · Cited by 43
Self-citationDevelopment of vision–based SLAM: from traditional methods to multimodal fusion
Zheng, Z., Su, K., Lin, S., Fu, Z., & Yang, C. (2024) · Robotic Intelligence and Automation · doi:10.1108/ria-10-2023-0142 · Cited by 29
Image features and seasons revisited
Krajník, T., Cristóforis, P. D., Nitsche, M., Kusumam, K., & Duckett, T. (2015) · European Conference on Mobile Robots · doi:10.1109/ecmr.2015.7324193 · Cited by 24
“Milford and Fraser [8] proposed to use image sequences to identify places across seasons, while Churchill and Newman [9] clustered different observations of the same place to form “experiences” that characterize the place appearance in particular conditions.”
BackgroundStone, T., Differt, D., Milford, M. J., & Webb, B. H. (2016) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2016.7487780 · Cited by 24
“In contrast, methods based on global features or whole images have proven to be highly invariant to substantial lighting changes, even when using low resolution imagery [4].”
BackgroundMethodInfluentialSelf-citationDeep learning features exception for cross-season visual place recognition
Kenshimov, C., Bampis, L., Amirgaliyev, B., Arslanov, M., & Gasteratos, A. C. (2017) · Pattern Recognition Letters · doi:10.1016/j.patrec.2017.10.028 · Cited by 24
Place recognition survey: An update on deep learning approachesPreprint
Barros, T., Pereira, R., Garrote, L., Premebida, C., & Nunes, U. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.10458 · Cited by 32
“…[157] Architecture: small Autoencoder L2 loss function Unsupervised feature embedding for visual loop closure Places [110]; KITTI [142]; Alderley [42]; Norland [2]; Gar-dens Point [22]; RGB + Thermal [88] Multispectral Domain Invariant model Architecture: CycleGAN [154] SSIM [155] + triplet +…”
BackgroundMethodInfluentialSeqSLAM with Bag of Visual Words for Appearance Based Loop Closure Detection
Tsintotas, K. A., Bampis, L., Rallis, S., & Gasteratos, A. C. (2018) · Mechanisms and machine science · doi:10.1007/978-3-030-00232-9_61 · Cited by 23
Advanced mapping robot and high-resolution dataset
Chen, H., Yang, Z., Zhao, X., Weng, G., Wan, H., Luo, J., Ye, X., Zhao, Z., He, Z., Shen, Y., et al. (2020) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2020.103559 · Cited by 25
Adversarial Feature Disentanglement for Place Recognition Across Changing Appearance
Tang, L., Wang, Y., Luo, Q., Ding, X., & Xiong, R. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196518 · Cited by 23
“We use Partitioned Nordland Dataset [34] and Alderley Day/Night Dataset [13] to validate our method in place recognition.”
MethodAccurate and Robust Visual Localization System in Large-Scale Appearance-Changing Environments
Yu, Y., Peng, Y., Xue, B., Jiao, J., Fan, R., & Liu, M. (2022) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2022.3177237 · Cited by 26
“To reduce false-positive matches and influence by appearance changes, Milford and Wyeth [16] matched a sequence of images rather than an individual image.”
BackgroundYe, T., Yan, X., Wang, S., Li, Y., & Zhou, F. (2022) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2022.3209727 · Cited by 23
“In recent years, vision-based place recognition methods [1], [2], [3] have been proposed; however, given that the visual sensor is highly sensitive to light, the visual-based approach is hugely unstable when applied in practice.”
BackgroundHow Many Events Do You Need? Event-Based Visual Place Recognition Using Sparse But Varying Pixels
Fischer, T., & Milford, M. J. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3216226 · Cited by 33
“As the observed number of events within a few selected pixels is relatively low, and thus the expressiveness of a single difference value is limited, we rely on sequence matching [28] to combine the temporal information contained in L query event frames.”
BackgroundMethodSelf-citation
Metadata
- Authors
- Milford, M. J., & Wyeth, G. F.
- Affiliations
- Queensland University of Technology (AU)
- Venue
- 2012 IEEE International Conference on Robotics and Automation
- Publisher
- IEEE
- Published
- 2012-05
- Pages
- 1643-1649
- ISSN
- 1050-4729 (ISSN-L)
- OpenAlex
- W2110405746
- Cited by
- 981
- References
- 22
- Citations by year
- 2026: 22 · 2025: 44 · 2024: 66 · 2023: 61 · 2022: 82 · 2021: 115 · 2020: 110 · 2019: 125 · 2018: 99 · 2017: 74 · 2016: 52 · 2015: 61 · 2014: 38 · 2013: 25 · 2012: 7
- FWCI
- 78.04
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Proceedings - IEEE International Conference on Robotics and Automation/Proceedings (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Robotics and Sensor-Based Localization
- Physical Sciences › Engineering › Automotive Engineering › Autonomous Vehicle Technology and Safety
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Keywords
- visual SLAM, visual navigation, route recognition, seasonal variation, heavy rainfall, vehicle-mounted camera, trajectory matching
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.