Pith. sign in

REVIEW 5 major objections 6 minor 84 references

Cross-Modal Geometric Hierarchy Fusion: An Implicit-Submap Driven Framework for Resilient 3D Place Recognition

T0 review · 5 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims a handcrafted 3D place-recognition descriptor built from an elastic-neural-point implicit submap outperforms scan-native and learned baselines on four LiDAR datasets while storing maps at about 1% of raw size.

desk verdict Good new idea — implicit submap as a place-recognition substrate — but the SOTA claim rests on a submap-only protocol that likely handicaps scan-native baselines; needs a neutrality check before I trust the margins. read the letter →

arxiv 2506.14243 v3 pith:FJSIKBCJ submitted 2025-06-17 cs.CV

Xiaohui Jiang , Haijiang Zhu , Chade Li , Fulin Tang , Ning An This is my paper · ORCID
classification cs.CV
keywords LiDARplacerecognitionimplicitneuralrepresentationelasticpointsloopclosuredetectionbird'seyeviewdescriptorsurfacenormalsubmapmapcompression
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

LiDAR place recognition typically breaks down when the same place is scanned at different densities and when a single geometric representation—raw points or a bird's-eye view—omits structure. This paper claims that converting each submap into an implicit, uniformly sampled surface field made of elastic neural points removes the density problem, and that deriving two complementary descriptors from that field—a macro bird's-eye-view layout descriptor and a micro surface-normal descriptor for dominant 3D segments—makes place recognition more discriminative than either view alone. Across KITTI, KITTI-360, NCLT, and MulRan, the fused descriptor reports the best AUC and max-F1 scores on nearly all compared short-term sequences (e.g., KITTI02 AUC 0.95 vs 0.85 for the best baseline), outperforms one learned baseline and trails another only slightly, runs in about 60 ms per submap, and compresses historical maps to roughly 1% of the raw cloud. If the claims hold, a training-free handcrafted descriptor can remain competitive with learned methods while doubling as a compact map representation.

What carries the argument

The load-bearing object is the elastic-neural-point implicit submap: a hash-indexed set of learned 3D points, each carrying a pose, latent feature, timestamps, and stability value, decoded by a shallow MLP into a signed distance field. This representation turns an arbitrary-density raw submap into a continuous, uniformly sampled surface, from which the pipeline extracts even mesh vertices, surface normals, and an occupancy grid. Those two outputs feed the two descriptor branches—a rotation-invariant bird's-eye-view descriptor built with log-Gabor filters, maximum-index maps, FAST keypoints, and VLAD, and a micro geometric descriptor built from spherical binning and pairwise angular differences of normals on dominant 3D segments—and the concatenation of the two branches is the final place-recognition vector.

What would settle it

Run all methods on a standard single-scan benchmark using their original input formats and released pipelines—no submap re-projection, no tensor shape adjustment—under the same 20 m ground-truth rule; if the reported AUC/F1 margins disappear or reverse, the advantage is an artifact of the submap-conversion and input-adaptation protocol rather than of the implicit descriptor itself.

Watch

Extended reading notes

Core claim

The central claim is that density-agnostic geometric reasoning, not richer learning, is what a robust 3D place-recognition descriptor needs. The paper builds an implicit submap from hash-indexed elastic neural points—each with position, orientation, latent feature, creation/update timestamps, and a stability score—decoded by a shallow MLP into signed distances; points whose SDF and stability exceed thresholds are dropped as dynamic. Marching cubes on the SDF field yields uniformly spaced mesh vertices with normals, and sign changes at voxel corners yield an occupancy grid. The occupancy grid is projected to a bird's-eye view and processed by log-Gabor filters, maximum-index maps, FAST keypoint detection, and VLAD pooling into a rotation-invariant macro descriptor; the normals on clustered 3D segments are binned on a sphere and their pairwise angular differences histogrammed into micro descriptors, with computation scheduled by how many BEV keypoints fall in a segment. Concatenating the macro and micro vectors gives the fused descriptor, which the paper reports surpasses the compared handcrafted baselines and one learning-based baseline while trailing another learning-based baseline slightly on KITTI02, and compresses KITTI02 from 9002.3 MB to 90.1 MB.

Load-bearing premise

The headline result depends on the comparison being fair to every baseline: scan-native descriptors were fed dense scans re-projected from submaps, pretrained learning-based models were given shape-adjusted inputs and not retrained, and the 20-meter loop-closure threshold was selected for urban scenes; if those choices handicap the baselines, the reported margins will not transfer.

Editorial extensions

If this is right

  • A single implicit submap can serve both as the source of the place-recognition descriptor and as the archived map, shrinking storage to about 1% of the raw cloud for the sequences tested.
  • The descriptor is handcrafted and label-free, so the same extraction code can transfer to a new sensor or environment without retraining, as long as the implicit-submap construction step is affordable.
  • Dynamic-object filtering before descriptor extraction should make repeated visits with moving cars and pedestrians more stable, because descriptors are built from the cleaned static surface field.
  • At roughly 60 ms per submap, the pipeline is fast enough for online loop-closure detection, although the paper does not integrate it into a full SLAM system.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper does not test cross-sensor retrieval—matching a submap built from a 64-line scan against one built from a 32-line scan. If the density-agnostic claim is right, the descriptor should hold up across sensors; that is a direct, untested consequence.
  • The two descriptor branches could be used as a coarse-to-fine cascade—BEV for candidate recall, normal-segment descriptor for geometric verification—rather than one concatenated vector; the paper never evaluates this cheaper variant.
  • The paper notes it has not been evaluated under a unified dataset benchmark, so its state-of-the-art claim is tied to its own protocol; published numbers from other papers are not directly comparable without re-running everything.
  • Runtime and memory are reported with a GPU helping to build the neural representation; a CPU-only deployment could shift the 60 ms figure substantially, which the paper does not measure.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. This manuscript proposes a LiDAR-based 3D place recognition framework built on an implicit neural point submap representation. The pipeline converts accumulated LiDAR frames into an elastic neural-point implicit surface, from which occupancy grids and surface normals are extracted. These feed two descriptor branches: a BEV descriptor based on log-Gabor filters and keypoint statistics, and a micro-geometry descriptor based on angular differences between normal vectors in clustered 3D segments; the two are concatenated into a fused descriptor. The authors evaluate on KITTI, KITTI-360, NCLT, and MulRan, comparing against handcrafted baselines and two learning-based methods, and report ablation, runtime, and storage results. The central claim is that the training-free handcrafted descriptor achieves state-of-the-art place recognition while compressing the stored map to roughly 1% of the raw cloud size.

Significance. If the experimental findings are substantiated, the contribution is significant: the paper proposes a genuinely different representation route for place recognition (implicit neural submaps feeding handcrafted descriptors), with an unusually broad multi-sensor evaluation and concrete storage/runtime advantages. The ablations in Table IV directionally support the design, and the cross-dataset coverage (four sensor types, short-term and long-term settings) is a strength. The claim of state-of-the-art performance, however, depends on an evaluation protocol that adapts scan-native and learned baselines to submaps rather than running them in their standard regimes, and several descriptor-defining parameters are left unspecified. These issues are load-bearing for the main claim and need to be resolved before the contribution can be assessed at face value.

major comments (5)
  1. [Section IV-A and IV-C3] The state-of-the-art claim ('Our method surpasses the compared approaches', Section IV-C) rests on an evaluation protocol that adapts baselines to submaps instead of testing them in their native setting. ScanContext++ and Ring++ are applied to submaps re-projected onto the middle frame, which the manuscript itself describes only as 'simulated a dense scan', and the learned baselines are given shape-adjusted multi-frame tensors without retraining. No single-scan unified benchmark is reported (the paper concedes this in Section V), so it is not established that the Table II margins (e.g., KITTI02 AUC 0.95 vs 0.85 for STD) transfer to conventional place-recognition settings. Please add a protocol-neutral evaluation: run the baselines on single scans with their standard input formats and standard KITTI-type loop-closure benchmarks, and/or evaluate the proposed descriptor on single scans, and report sensitivity of the results to the 20 m positive-pair threshold.
  2. [Section III-C, III-D, and Table I] The final descriptor is underspecified. The geometric descriptor FG is stated to lie in Ra and the BEV descriptor FB in Rb, but the dimensions a and b are never given. The VLAD codebook size is said to be set 'proportionally to Nc' during encoding, but no formula or exact mapping is provided. The log-Gabor filter parameters fs, σf, σω, and the number of directions No in Eqs. (6)-(11) are not listed in Table I. These quantities determine descriptor dimensionality and discriminative power; without them the method cannot be reproduced and the comparison tables cannot be independently interpreted. Please specify all such parameters explicitly.
  3. [Section IV-C3] The learning-based comparison is too narrow to support the broad state-of-the-art claim. Only one sequence per baseline is used (BEVPlace++ on KITTI-02 and LCDNet on KITTI-360-00), the tensor-shape adjustment is made without retraining, and the manuscript reports being 'marginally behind BEVPlace++' on KITTI-02 at higher recall. This does not establish superiority over learning-based methods. Please either expand the learned-baseline comparison to additional sequences with a clear description of how the pretrained weights interact with the adjusted input tensors, or temper the claim to 'competitive with' rather than 'surpasses' these methods.
  4. [Section IV-B and Tables II-III] The positive-pair criterion uses average submap positions, an index difference greater than 50, and a 20 m distance threshold that was 'chosen based on the urban environment'. The same criterion is used both to define ground-truth positive pairs and to select per-method decision thresholds via max F1, yet no analysis of threshold sensitivity is provided. Since the claimed margins over baselines are the central result, please report at least one alternative distance threshold (e.g., 5 m and 10 m) and describe how the decision threshold for each method was selected, to show that the reported differences are not artifacts of this particular criterion.
  5. [Section III-B and Abstract] The method is described as 'training-free' in the Abstract, Introduction, and Section IV-C3, but the implicit representation includes a neural decoder Dθ (Section III-B2) that must be optimized on each submap. The per-submap optimization procedure (number of iterations, loss function, learning schedule, number of LiDAR frames accumulated) is not described, and it is not stated whether any pretraining of the decoder is used. This matters because the runtime and storage comparisons in Tables V-VI include the cost of building this representation. Please clarify exactly what is trained per submap and how the 'training-free' characterization is intended.
minor comments (6)
  1. [Section III-D, Eq. (9)] Equation (9) defines the MIM as arg max over o of A(ρ, θ, 0), which appears to be a typo for A(ρ, θ, o). Also in Eq. (6) the symbol ω0 is used in the denominator but the text refers to ωo.
  2. [Section IV-A] The text says 'we selected five handcrafted feature extraction methods for comparison' but then lists six methods: Scan Context++, M2DP, NDT, BoW3D, Ring++, and STD. Please correct the count.
  3. [Table VI] For NCLT02, the reported memory is 108.2 MB with the annotation '0.07%' of 15366 MB, but 108.2/15366 is approximately 0.7%. Please correct the percentage.
  4. [Fig. 5 caption] The caption states that 'all image coordinates (x, y) are normalized to the unit interval [0, 1]', but the figure shows precision-recall curves. The caption should refer to the axes of the PR plots rather than image coordinates.
  5. [Section V] The conclusion states that the method 'has not been evaluated under a unified dataset benchmark', which is in tension with the abstract's claim of 'extensive experiments' on four datasets. Please clarify the scope of the limitation and how the current evaluations relate to standard benchmarks.
  6. [Table IV] The ablations label rows only with checkmarks, making it difficult to see which component is removed in each row. Please use explicit component names or descriptions (for example, 'w/o normal vector') to improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the descriptor pipeline is an empirical construction, and the reported results are measurements rather than fitted predictions; the acknowledged lack of a unified benchmark is an evaluation-validity concern, not a circularity.

full rationale

The paper does not present a derivation in which an output is defined in terms of its own target. The fused descriptor is constructed from occupancy grids and surface normals derived from an implicit submap representation (Sections III-C and III-D), and performance is then measured on standard datasets. No parameter is fitted to the reported AUC/F1 values and then renamed as a prediction: Table I lists fixed hyperparameters, and Section IV-B defines the positive-pair criterion by a 20 m distance threshold chosen on the basis of urban environment and LiDAR range, not fitted to the test outputs. The comparison with ScanContext++ and Ring++ re-projects submaps onto a middle frame, which the paper itself describes as 'simulated a dense scan' (Section IV-A), and learning-based baselines are given shape-adjusted tensors without retraining (Section IV-C3); these are legitimate protocol-validity concerns but not circularity, because the baseline scores are not used as inputs to the proposed method. The paper also openly concedes in Section V that 'the current method has not been evaluated under a unified dataset benchmark', which weakens the generality of the SOTA claim but does not make the claim self-referential. Citations to prior work, including the elastic-neural-point representation from PIN-SLAM [47], are standard literature support rather than load-bearing self-citations, and none of the authors' own prior results is invoked to justify the central claim. The derivation chain is therefore self-contained as an engineering construction, and no equation-level reduction of a prediction to its input was found.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a modest set of imported machinery and hand-set hyperparameters: the elastic-neural-point mapping is taken from PIN-SLAM [47] without restating its optimization objective; the descriptor modules (log-Gabor, MIM/FAST, VLAD, FEC) are all prior published components; and the pipeline has no learned parameters for the descriptor itself. The load-bearing choices are therefore the 12-plus hyperparameters in Table I, several of which (submap length, voxel size, filter thresholds) interact directly with the evaluation protocol used to compute the reported AUC/F1 numbers, and the descriptor dimensions a and b are not even disclosed. No new physical or algorithmic entity is introduced; the 'elastic neural points' entity is adopted from prior work, which is why the invented-entities list is empty.

free parameters (6)
  • Submap length tau_n = 25 m
    Defines how much driving distance is accumulated into one submap and therefore one descriptor (Table I). It interacts with the 20 m positive-pair threshold in Section IV-B, so submap size and the ground-truth definition are coupled; no sensitivity analysis is given.
  • Voxel resolution r = 0.2 m
    Sets the SDF query grid, the marching-cubes reconstruction scale, the occupancy classification, and the BEV pixel physical size (Table I). It directly controls how much geometric detail survives into both descriptor branches.
  • Dynamic-object thresholds gamma_d, gamma_mu = 0.006 rmax, 4
    Ad hoc thresholds (Table I) that decide when a sampled point is dynamic and filtered (Section III-B2). Wrong values would either keep moving objects in the map or erase static structure; no sensitivity study is provided.
  • Descriptor design parameters = 72 spherical bins, 18 histogram dims, nm=400, J=96, l=6, Ns=4
    Hand-chosen binning, sampling, patch, and scale counts for the two descriptor branches (Section III-C and III-D, Table I). These fix the shape of the features and are never ablated individually.
  • Final descriptor dimensions a and b = unspecified
    Section III-C defines FG in R^a and Section III-D defines FB in R^b, with the VLAD vector zero-padded to length a, but neither a nor b, nor the K proportional to Nc codebook rule, is given anywhere in the paper. The final feature dimensionality is a free parameter the reader cannot recover.
  • Loop-closure ground-truth thresholds = 20 m distance, index gap 50
    Section IV-B defines positive pairs by submap-position distance at most 20 m and index difference greater than 50, 'chosen based on the urban environment' and LiDAR range. These thresholds determine every TP/FP/FN label behind the reported AUC and F1 numbers, and are chosen on the same datasets that are scored.
assumptions (5)
  • domain assumption The elastic-neural-point SDF representation from PIN-SLAM [47] produces globally consistent, uniformly distributed implicit geometry when applied per-submap with the Table I hyperparameters.
    The paper imports the representation and its optimization without restating the loss or stability update (Section III-B references the construction but not the training objective). If per-submap optimization fails to converge or leaves holes in cluttered scenes, the normals and occupancy grids degrade.
  • domain assumption A uniform distribution of neural points implies descriptor invariance to input point cloud density.
    The abstract and Section III-B assert the representation is 'immune to the interference of original scene point cloud density'. This property is asserted, not derived; end-to-end density invariance is only indirectly tested by using different sensors on different datasets.
  • domain assumption Stability-and-SDF thresholding removes dynamic objects without damaging static structure.
    Section III-B2 classifies a point as dynamic when S(pW) > gamma_d and H(pW) > gamma_mu. Fixed thresholds assume dynamic objects are consistently unstable; false removals and missed removals are not measured.
  • standard math Marching cubes on a 0.2 m voxel grid yields surface vertices and normals accurate enough for normal-angle histograms.
    Marching cubes is standard, but its fidelity at this resolution on 25 m outdoor submaps is unvalidated: no comparison of reconstructed normals against ground-truth surfaces is given.
  • domain assumption Image-based tools (log-Gabor filters, MIM, FAST, VLAD) transfer without modification to BEV occupancy projections.
    Section III-D applies RIFT-style image matching tools [75], [76] to BEV images of occupancy grids; the transfer is claimed to work, not analyzed, and the key filter parameters (fs, sigma_f, sigma_omega, number of directions) are never listed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Cross-Modal Geometric Hierarchy Fusion: An Implicit-Submap Driven Framework for Resilient 3D Place Recognition." pith.science (2026). https://pith.science/paper/FJSIKBCJ

@misc{pith2026250614243,
  author       = {Pith},
  title        = {Pith review of: Cross-Modal Geometric Hierarchy Fusion: An Implicit-Submap Driven Framework for Resilient 3D Place Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FJSIKBCJ}},
  note         = {Machine review of arXiv:2506.14243}
}
read the original abstract

LiDAR-based place recognition serves as a crucial enabler for long-term autonomy in robotics and autonomous driving systems. Yet, prevailing methodologies relying on handcrafted feature extraction face dual challenges: (1) Inconsistent point cloud density, induced by ego-motion dynamics and environmental disturbances during repeated traversals, leads to descriptor instability, and (2) Representation fragility stems from reliance on single-level geometric abstractions that lack discriminative power in structurally complex scenarios. To address these limitations, we propose a novel framework that redefines 3D place recognition through density-agnostic geometric reasoning. Specifically, we introduce an implicit 3D representation based on elastic points, which is immune to the interference of original scene point cloud density and achieves the characteristic of uniform distribution. Subsequently, we derive the occupancy grid and normal vector information of the scene from this implicit representation. Finally, with the aid of these two types of information, we obtain descriptors that fuse geometric information from both bird's-eye view (capturing macro-level spatial layouts) and 3D segment (encoding micro-scale surface geometries) perspectives. We conducted extensive experiments on numerous datasets (KITTI, KITTI-360, MulRan, NCLT) across diverse environments. The experimental results demonstrate that our method achieves state-of-the-art performance. Moreover, our approach strikes an optimal balance between accuracy, runtime, and memory optimization for historical maps, showcasing excellent Resilient and scalability. Our code will be open-sourced in the future.

Figures

Figures reproduced from arXiv: 2506.14243 by the authors.

Figure 1
Figure 1. Our 3D place recognition pipeline, unlike other methods, first converts [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. To better handle the random noise caused by variations [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 2
Figure 2. This is the pipeline for our 3D Place Recognition. We feed a specific number of LiDAR frames to form a sub-map, which we then transform into an [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figures from the paper (7 more)
Figure 3
Figure 3. Figure 3: (a) The figure illustrates the extraction of geometric descriptors for [PITH_FULL_IMAGE:figures/full_fig_p006_3.png]
Figure 4
Figure 4. Figure 4: Datasets for experiment. A. Dataset and Experimental Settings Below, we provide a detailed description of the dataset used in this study. Our dataset includes various types of LiDAR and different application scenarios. The configuration of the LiDAR sensors used in thi…
Figure 5
Figure 5. Figure 5: Evaluation of twenty short-term sequences reveals that each subfigure in Figure 5 corresponds to the precision-recall performance of a specific [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Loop retrieval result using each method’s max F1 score threshold on KITTI02. [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Loop retrieval result using each method’s max F1 score threshold on MulRan04. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 9
Figure 9. Figure 9: Precision–recall curve KITTI02, KITTI360-00 on learning-based [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 8
Figure 8. Figure 8: Precision–recall curve NCLT, MulRan on long-term. [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

84 extracted references · 78 canonical work pages

  1. [1]

    3D LiDAR-Based Place Recognition Tech- niques: A Review of the Past 10 Years[J]

    Du Z, Ji S, Khoshelham K. 3D LiDAR-Based Place Recognition Tech- niques: A Review of the Past 10 Years[J]. IEEE Transactions on Instru- mentation and Measurement, 2024

    2024

  2. [2]

    Global localization in repetitive and ambiguous environments[C]//2023 IEEE International Conference on Robotics and Automation (ICRA)

    Wu Z, Wang W, Zhang J, et al. Global localization in repetitive and ambiguous environments[C]//2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023: 12374-12380

    2023

  3. [3]

    Loopy-slam: Dense neural slam with loop closures[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Liso L, Sandstr ¨om E, Yugay V , et al. Loopy-slam: Dense neural slam with loop closures[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024: 20363-20373

    2024

  4. [4]

    Quatro++: Robust global registration exploiting ground segmentation for loop closing in LiDAR SLAM[J]

    Lim H, Kim B, Kim D, et al. Quatro++: Robust global registration exploiting ground segmentation for loop closing in LiDAR SLAM[J]. The International Journal of Robotics Research, 2024, 43(5): 685-715

    2024

  5. [5]

    Dlc-slam: A robust lidar-slam system with learning-based denoising and loop closure[J]

    Liu K, Cao M. Dlc-slam: A robust lidar-slam system with learning-based denoising and loop closure[J]. IEEE/ASME Transactions on Mechatron- ics, 2023, 28(5): 2876-2884

    work page 2023

  6. [6]

    A review on map-merging methods for typical map types in multiple-ground-robot SLAM solutions[J]

    Yu S, Fu C, Gostar A K, et al.. A review on map-merging methods for typical map types in multiple-ground-robot SLAM solutions[J]. Sensors, 2020, 20(23): 6988

    work page 2020

  7. [7]

    Place recognition and navigation of outdoor mobile robots based on random Forest learning with a 3D LiDAR[J]

    Zhou B, He Y , Huang W, et al. Place recognition and navigation of outdoor mobile robots based on random Forest learning with a 3D LiDAR[J]. Journal of Intelligent and Robotic Systems, 2022, 104(4): 72

    work page 2022

  8. [8]

    Placenav: Topological navigation through place recognition[C]//2024 IEEE International Conference on Robotics and Automation (ICRA)

    Suomela L, Kalliola J, Edelman H, et al. Placenav: Topological navigation through place recognition[C]//2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024: 5205-5213

    work page 2024

Show all 84 references
  1. [9]

    Effectively detecting loop closures using point cloud density maps[C]//2024 IEEE International Conference on Robotics and Automation (ICRA)

    Gupta S, Guadagnino T, Mersch B, et al. Effectively detecting loop closures using point cloud density maps[C]//2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024: 10260- 10266

    2024

  2. [10]

    Pyramid point cloud transformer for large-scale place recognition[C]//Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision

    Hui L, Yang H, Cheng M, et al. Pyramid point cloud transformer for large-scale place recognition[C]//Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision. 2021: 6098-6107

    2021

  3. [12]

    Robust place recognition using an imaging lidar[C]//2021 IEEE international conference on robotics and automation (ICRA)

    Shan T, Englot B, Duarte F, et al. Robust place recognition using an imaging lidar[C]//2021 IEEE international conference on robotics and automation (ICRA). IEEE, 2021: 5469-5475. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 13

    2021

  4. [13]

    Contour context: Abstract structural distribution for 3d lidar loop detection and metric pose estimation[C]//2023 IEEE in- ternational conference on robotics and automation (ICRA)

    Jiang B, Shen S. Contour context: Abstract structural distribution for 3d lidar loop detection and metric pose estimation[C]//2023 IEEE in- ternational conference on robotics and automation (ICRA). IEEE, 2023: 8386-8392

    2023

  5. [14]

    Osk: A novel lidar occupancy set key- based place recognition method in urban environment[J]

    Zhang Z, Huang Y , Si S, et al. Osk: A novel lidar occupancy set key- based place recognition method in urban environment[J]. IEEE Transac- tions on Instrumentation and Measurement, 2024

    2024

  6. [15]

    SegMap: Segment-based mapping and localization using data-driven descriptors[J]

    Dube R, Cramariuc A, Dugas D, et al. SegMap: Segment-based mapping and localization using data-driven descriptors[J]. The International Journal of Robotics Research, 2020, 39(2-3): 339-355

    2020

  7. [16]

    Adafusion: Visual-lidar fusion with adaptive weights for place recognition[J]

    Lai H, Yin P, Scherer S. Adafusion: Visual-lidar fusion with adaptive weights for place recognition[J]. IEEE Robotics and Automation Letters, 2022, 7(4): 12038-12045

    2022

  8. [17]

    Pointnetvlad: Deep point cloud based retrieval for large-scale place recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition

    Uy M A, Lee G H. Pointnetvlad: Deep point cloud based retrieval for large-scale place recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 4470-4479

    2018

  9. [18]

    Std: Stable triangle descriptor for 3d place recognition[C]//2023 IEEE international conference on robotics and automation (ICRA)

    Yuan C, Lin J, Zou Z, et al. Std: Stable triangle descriptor for 3d place recognition[C]//2023 IEEE international conference on robotics and automation (ICRA). IEEE, 2023: 1897-1903

    2023

  10. [19]

    Local descriptor for robust place recognition using lidar intensity[J]

    Guo J, Borges P V K, Park C, et al. Local descriptor for robust place recognition using lidar intensity[J]. IEEE Robotics and Automation Letters, 2019, 4(2): 1470-1477

    2019

  11. [20]

    LoGG3D-Net: Lo- cally guided global descriptor learning for 3D place recognition[C]//2022 International Conference on Robotics and Automation (ICRA)

    Vidanapathirana K, Ramezani M, Moghadam P, et al. LoGG3D-Net: Lo- cally guided global descriptor learning for 3D place recognition[C]//2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022: 2215-2221

    2022

  12. [21]

    BEVPlace: Learning LiDAR-based place recognition using bird’s eye view images[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision

    Luo L, Zheng S, Li Y , et al. BEVPlace: Learning LiDAR-based place recognition using bird’s eye view images[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. 2023: 8700- 8709

    2023

  13. [22]

    A heterogeneous 3D map-based place recognition solution using virtual LiDAR and a polar grid height coding image descriptor[J]

    Xu D, Liu J, Hyypp ¨a J, et al. A heterogeneous 3D map-based place recognition solution using virtual LiDAR and a polar grid height coding image descriptor[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2022, 183: 1-18

    2022

  14. [23]

    Visual place recognition using lidar intensity information[C]//2021 IEEE/RSJ International Confer- ence on Intelligent Robots and Systems (IROS)

    Di Giammarino L, Aloise I, Stachniss C, et al. Visual place recognition using lidar intensity information[C]//2021 IEEE/RSJ International Confer- ence on Intelligent Robots and Systems (IROS). IEEE, 2021: 4382-4389

    2021

  15. [24]

    V oxel-based representation learning for place recognition based on 3d point clouds[C]//2020 IEEE/RSJ Interna- tional Conference on Intelligent Robots and Systems (IROS)

    Siva S, Nahman Z, Zhang H. V oxel-based representation learning for place recognition based on 3d point clouds[C]//2020 IEEE/RSJ Interna- tional Conference on Intelligent Robots and Systems (IROS). IEEE, 2020: 8351-8357

    2020

  16. [25]

    Feature Extraction of Horizontal Plane and Optimization of 3D LiDAR SLAM in Indoor Environments[J]

    Wang S, Li H, Miao T, et al. Feature Extraction of Horizontal Plane and Optimization of 3D LiDAR SLAM in Indoor Environments[J]. IEEE Transactions on Instrumentation and Measurement, 2025

    2025

  17. [26]

    LE-VINS: A robust solid-state-LiDAR- enhanced visual-inertial navigation system for low-speed robots[J]

    Tang H, Niu X, Zhang T, et al. LE-VINS: A robust solid-state-LiDAR- enhanced visual-inertial navigation system for low-speed robots[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 1-13

    2023

  18. [27]

    A comparative analysis of LiDAR SLAM- based indoor navigation for autonomous vehicles[J]

    Zou Q, Sun Q, Chen L, et al. A comparative analysis of LiDAR SLAM- based indoor navigation for autonomous vehicles[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(7): 6907-6921

    2021

  19. [28]

    Slamesh: Real-time lidar simultaneous localization and meshing[C]//2023 IEEE International Conference on Robotics and Automation (ICRA)

    Ruan J, Li B, Wang Y , et al. Slamesh: Real-time lidar simultaneous localization and meshing[C]//2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023: 3546-3552

    2023

  20. [29]

    V oxel-Based Navigation: A Systematic Re- view of Techniques, Applications, and Challenges[J]

    Niu L, Wang Z, Lin Z, et al. V oxel-Based Navigation: A Systematic Re- view of Techniques, Applications, and Challenges[J]. ISPRS International Journal of Geo-Information, 2024, 13(12): 461

    2024

  21. [30]

    Ren Y , Cai Y , Zhu F, et al. Rog-map: An efficient robocentric occupancy grid map for large-scene and high-resolution lidar-based motion plan- ning[C]//2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2024: 8119-8125

    2024

  22. [31]

    Nerf: Representing scenes as neural radiance fields for view synthesis[J]

    Mildenhall B, Srinivasan P P, Tancik M, et al. Nerf: Representing scenes as neural radiance fields for view synthesis[J]. Communications of the ACM, 2021, 65(1): 99-106

    2021

  23. [32]

    Occupancy networks: Learning 3d reconstruction in function space[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Mescheder L, Oechsle M, Niemeyer M, et al. Occupancy networks: Learning 3d reconstruction in function space[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 4460-4470

    2019

  24. [33]

    Deepsdf: Learning continuous signed distance functions for shape representation[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Park J J, Florence P, Straub J, et al. Deepsdf: Learning continuous signed distance functions for shape representation[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 165-174

    2019

  25. [34]

    Point-nerf: Point-based neural radiance fields[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Xu Q, Xu Z, Philip J, et al. Point-nerf: Point-based neural radiance fields[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2022: 5438-5448

    2022

  26. [35]

    Instant neural graphics primitives with a multiresolution hash encoding[J]

    M ¨uller T, Evans A, Schied C, et al. Instant neural graphics primitives with a multiresolution hash encoding[J]. ACM transactions on graphics (TOG), 2022, 41(4): 1-15

    2022

  27. [36]

    Neural geometric level of de- tail: Real-time rendering with implicit 3d shapes[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Takikawa T, Litalien J, Yin K, et al. Neural geometric level of de- tail: Real-time rendering with implicit 3d shapes[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021: 11358-11367

    2021

  28. [37]

    Neural rgb-d surface reconstruction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Azinovi ´c D, Martin-Brualla R, Goldman D B, et al. Neural rgb-d surface reconstruction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022: 6290-6301

    2022

  29. [38]

    imap: Implicit mapping and positioning in real-time[C]//Proceedings of the IEEE/CVF international conference on computer vision

    Sucar E, Liu S, Ortiz J, et al. imap: Implicit mapping and positioning in real-time[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2021: 6229-6238

    2021

  30. [39]

    Eslam: Efficient dense slam system based on hybrid representation of signed distance fields[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion

    Johari M M, Carta C, Fleuret F. Eslam: Efficient dense slam system based on hybrid representation of signed distance fields[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion. 2023: 17408-17419

    2023

  31. [40]

    Co-slam: Joint coordinate and sparse parametric encodings for neural real-time slam[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Wang H, Wang J, Agapito L. Co-slam: Joint coordinate and sparse parametric encodings for neural real-time slam[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2023: 13293-13302

    2023

  32. [41]

    Nice-slam: Neural implicit scalable encoding for slam[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Zhu Z, Peng S, Larsson V , et al. Nice-slam: Neural implicit scalable encoding for slam[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2022: 12786-12796

    2022

  33. [42]

    Ir-mcl: Implicit representation- based online global localization[J]

    Kuang H, Chen X, Guadagnino T, et al. Ir-mcl: Implicit representation- based online global localization[J]. IEEE Robotics and Automation Let- ters, 2023, 8(3): 1627-1634

    2023

  34. [43]

    Locndf: Neural distance field mapping for robot localization[J]

    Wiesmann L, Guadagnino T, Vizzo I, et al. Locndf: Neural distance field mapping for robot localization[J]. IEEE Robotics and Automation Letters, 2023, 8(8): 4999-5006

    2023

  35. [44]

    Shine-mapping: Large-scale 3d mapping using sparse hierarchical implicit neural representations[C]//2023 IEEE International Conference on Robotics and Automation (ICRA)

    Zhong X, Pan Y , Behley J, et al. Shine-mapping: Large-scale 3d mapping using sparse hierarchical implicit neural representations[C]//2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023: 8371-8377

    2023

  36. [45]

    Loner: Lidar only neural representations for real-time slam[J]

    Isaacson S, Kung P C, Ramanagopal M, et al. Loner: Lidar only neural representations for real-time slam[J]. IEEE Robotics and Automation Letters, 2023, 8(12): 8042-8049

    2023

  37. [46]

    Nerf-loam: Neural implicit representation for large-scale incremental lidar odometry and mapping[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision

    Deng J, Wu Q, Chen X, et al. Nerf-loam: Neural implicit representation for large-scale incremental lidar odometry and mapping[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. 2023: 8218-8227

    2023

  38. [47]

    PIN-SLAM: LiDAR SLAM using a point-based implicit neural representation for achieving global map consistency[J]

    Pan Y , Zhong X, Wiesmann L, et al. PIN-SLAM: LiDAR SLAM using a point-based implicit neural representation for achieving global map consistency[J]. IEEE Transactions on Robotics, 2024

    2024

  39. [48]

    Aligning point cloud views using persistent feature histograms[C]//2008 IEEE/RSJ international con- ference on intelligent robots and systems

    Rusu R B, Blodow N, Marton Z C, et al. Aligning point cloud views using persistent feature histograms[C]//2008 IEEE/RSJ international con- ference on intelligent robots and systems. IEEE, 2008: 3384-3391

    2008

  40. [49]

    Fast point feature histograms (FPFH) for 3D registration[C]//2009 IEEE international conference on robotics and automation

    Rusu R B, Blodow N, Beetz M. Fast point feature histograms (FPFH) for 3D registration[C]//2009 IEEE international conference on robotics and automation. IEEE, 2009: 3212-3217

    2009

  41. [50]

    SHOT: Unique signatures of histograms for surface and texture description[J]

    Salti S, Tombari F, Di Stefano L. SHOT: Unique signatures of histograms for surface and texture description[J]. Computer vision and image under- standing, 2014, 125: 251-264

    2014

  42. [51]

    Using spin images for efficient object recog- nition in cluttered 3D scenes[J]

    Johnson A E, Hebert M. Using spin images for efficient object recog- nition in cluttered 3D scenes[J]. IEEE Transactions on pattern analysis and machine intelligence, 1999, 21(5): 433-449

    1999

  43. [52]

    Loop closure detection using small-sized signatures from 3D LIDAR data[C]//2011 IEEE International Symposium on Safety, Security, and Rescue Robotics

    Muhammad N, Lacroix S. Loop closure detection using small-sized signatures from 3D LIDAR data[C]//2011 IEEE International Symposium on Safety, Security, and Rescue Robotics. IEEE, 2011: 333-338

    2011

  44. [53]

    A fast histogram-based similarity measure for detecting loop closures in 3-d lidar data[C]//2015 IEEE/RSJ interna- tional conference on intelligent robots and systems (IROS)

    R ¨ohling T, Mack J, Schulz D. A fast histogram-based similarity measure for detecting loop closures in 3-d lidar data[C]//2015 IEEE/RSJ interna- tional conference on intelligent robots and systems (IROS). IEEE, 2015: 736-741

    2015

  45. [54]

    M2DP: A novel 3D point cloud descriptor and its application in loop closure detection[C]//2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

    He L, Wang X, Zhang H. M2DP: A novel 3D point cloud descriptor and its application in loop closure detection[C]//2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2016: 231- 237

    2016

  46. [55]

    Scan context: Egocentric spatial descriptor for place recognition within 3d point cloud map[C]//2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

    Kim G, Kim A. Scan context: Egocentric spatial descriptor for place recognition within 3d point cloud map[C]//2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2018: 4802- 4809

    2018

  47. [56]

    Scan context++: Structural place recognition robust to rotation and lateral variations in urban environments[J]

    Kim G, Choi S, Kim A. Scan context++: Structural place recognition robust to rotation and lateral variations in urban environments[J]. IEEE Transactions on Robotics, 2021, 38(3): 1856-1874

    2021

  48. [57]

    Intensity scan context: Coding intensity and geometry relations for loop closure detection[C]//2020 IEEE international conference on robotics and automation (ICRA)

    Wang H, Wang C, Xie L. Intensity scan context: Coding intensity and geometry relations for loop closure detection[C]//2020 IEEE international conference on robotics and automation (ICRA). IEEE, 2020: 2095-2101

    2020

  49. [58]

    Place recognition of large-scale unstructured orchards with attention score maps[J]

    Ou F, Li Y , Miao Z. Place recognition of large-scale unstructured orchards with attention score maps[J]. IEEE Robotics and Automation Letters, 2023, 8(2): 958-965. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    2023

  50. [59]

    Lidar iris for loop-closure detec- tion[C]//2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

    Wang Y , Sun Z, Xu C Z, et al. Lidar iris for loop-closure detec- tion[C]//2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2020: 5769-5775

    2020

  51. [60]

    Pointnet: Deep learning on point sets for 3d classification and segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition

    Qi C R, Su H, Mo K, et al. Pointnet: Deep learning on point sets for 3d classification and segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 652-660

    2017

  52. [61]

    NetVLAD: CNN architecture for weakly supervised place recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition

    Arandjelovic R, Gronat P, Torii A, et al. NetVLAD: CNN architecture for weakly supervised place recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 5297-5307

    2016

  53. [62]

    Xia Y , Xu Y , Li S, et al. SOE-Net: A self-attention and orientation encoding network for point cloud based place recognition[C]//Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition. 2021: 11348-11357

    2021

  54. [63]

    PCAN: 3D attention map learning using contex- tual information for point cloud based retrieval[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Zhang W, Xiao C. PCAN: 3D attention map learning using contex- tual information for point cloud based retrieval[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 12436-12445

    2019

  55. [64]

    Locus: Lidar-based place recognition using spatiotemporal higher-order pooling[C]//2021 IEEE International Conference on Robotics and Automation (ICRA)

    Vidanapathirana K, Moghadam P, Harwood B, et al. Locus: Lidar-based place recognition using spatiotemporal higher-order pooling[C]//2021 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2021: 5075-5081

    2021

  56. [65]

    Hitpr: Hierarchical transformer for place recognition in point cloud[C]//2022 International Conference on Robotics and Automation (ICRA)

    Hou Z, Yan Y , Xu C, et al. Hitpr: Hierarchical transformer for place recognition in point cloud[C]//2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022: 2612-2618

    2022

  57. [66]

    Attention is all you need[J]

    Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30

    2017

  58. [67]

    1-day learning, 1-year localization: Long- term lidar localization using scan context image[J]

    Kim G, Park B, Kim A. 1-day learning, 1-year localization: Long- term lidar localization using scan context image[J]. IEEE Robotics and Automation Letters, 2019, 4(2): 1948-1955

    2019

  59. [68]

    SSC: Semantic scan context for large- scale place recognition[C]//2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

    Li L, Kong X, Zhao X, et al. SSC: Semantic scan context for large- scale place recognition[C]//2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2021: 2092-2099

    2021

  60. [69]

    Lpd-net: 3d point cloud learning for large-scale place recognition and environment analysis[C]//Proceedings of the IEEE/CVF international conference on computer vision

    Liu Z, Zhou S, Suo C, et al. Lpd-net: 3d point cloud learning for large-scale place recognition and environment analysis[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2019: 2831- 2840

    2019

  61. [70]

    A registration-aided domain adaptation network for 3D point cloud based place recognition[C]//2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

    Qiao Z, Hu H, Shi W, et al. A registration-aided domain adaptation network for 3D point cloud based place recognition[C]//2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2021: 1317-1322

    2021

  62. [71]

    Density-Driven Adaptive Hybrid Network for Large-Scale Place Recognition[J]

    Wang S, Zhang Y , Zhang J, et al. Density-Driven Adaptive Hybrid Network for Large-Scale Place Recognition[J]. IEEE Transactions on Instrumentation and Measurement, 2025

    2025

  63. [72]

    Fast segmentation of 3D point clouds: A paradigm on LiDAR data for autonomous vehicle applications[C]//2017 IEEE International Conference on Robotics and Automation (ICRA)

    Zermas D, Izzat I, Papanikolopoulos N. Fast segmentation of 3D point clouds: A paradigm on LiDAR data for autonomous vehicle applications[C]//2017 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2017: 5067-5073

    2017

  64. [73]

    FEC: Fast Euclidean clustering for point cloud segmentation[J]

    Cao Y , Wang Y , Xue Y , et al. FEC: Fast Euclidean clustering for point cloud segmentation[J]. Drones, 2022, 6(11): 325

    2022

  65. [74]

    Aggregating local descriptors into a compact image representation[C]//2010 IEEE computer society conference on computer vision and pattern recognition

    J ´egou H, Douze M, Schmid C, et al. Aggregating local descriptors into a compact image representation[C]//2010 IEEE computer society conference on computer vision and pattern recognition. IEEE, 2010: 3304-3311

    2010

  66. [75]

    Self-invertible 2D log-Gabor wavelets[J]

    Fischer S, ˇSroubek F, Perrinet L, et al. Self-invertible 2D log-Gabor wavelets[J]. International Journal of Computer Vision, 2007, 75: 231-246

    2007

  67. [76]

    RIFT: Multi-modal image matching based on radiation-variation insensitive feature transform[J]

    Li J, Hu Q, Ai M. RIFT: Multi-modal image matching based on radiation-variation insensitive feature transform[J]. IEEE Transactions on Image Processing, 2019, 29: 3296-3310

    2019

  68. [77]

    Are we ready for autonomous driving? the kitti vision benchmark suite[C]//2012 IEEE conference on computer vision and pattern recognition

    Geiger A, Lenz P, Urtasun R. Are we ready for autonomous driving? the kitti vision benchmark suite[C]//2012 IEEE conference on computer vision and pattern recognition. IEEE, 2012: 3354-3361

    2012

  69. [78]

    Kitti-360: A novel dataset and benchmarks for urban scene understanding in 2d and 3d[J]

    Liao Y , Xie J, Geiger A. Kitti-360: A novel dataset and benchmarks for urban scene understanding in 2d and 3d[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 45(3): 3292-3310

    2022

  70. [79]

    University of Michigan North Campus long-term vision and lidar dataset[J]

    Carlevaris-Bianco N, Ushani A K, Eustice R M. University of Michigan North Campus long-term vision and lidar dataset[J]. The International Journal of Robotics Research, 2016, 35(9): 1023-1035

    2016

  71. [80]

    Mulran: Multimodal range dataset for urban place recognition[C]//2020 IEEE international conference on robotics and automation (ICRA)

    Kim G, Park Y S, Cho Y , et al. Mulran: Multimodal range dataset for urban place recognition[C]//2020 IEEE international conference on robotics and automation (ICRA). IEEE, 2020: 6246-6253

    2020

  72. [81]

    Automatic appearance- based loop detection from three-dimensional laser data using the normal distributions transform[J]

    Magnusson M, Andreasson H, N ¨uchter A, et al. Automatic appearance- based loop detection from three-dimensional laser data using the normal distributions transform[J]. Journal of Field Robotics, 2009, 26(11-12): 892-914

    2009

  73. [82]

    Bow3d: Bag of words for real-time loop closing in 3d lidar slam[J]

    Cui Y , Chen X, Zhang Y , et al. Bow3d: Bag of words for real-time loop closing in 3d lidar slam[J]. IEEE Robotics and Automation Letters, 2022, 8(5): 2828-2835

    2022

  74. [83]

    Ring++: Roto-translation invariant gram for global localization on a sparse scan map[J]

    Xu X, Lu S, Wu J, et al. Ring++: Roto-translation invariant gram for global localization on a sparse scan map[J]. IEEE Transactions on Robotics, 2023, 39(6): 4616-4635

    2023

  75. [84]

    Lcdnet: Deep loop closure detection and point cloud registration for lidar slam[J]

    Cattaneo D, Vaghi M, Valada A. Lcdnet: Deep loop closure detection and point cloud registration for lidar slam[J]. IEEE Transactions on Robotics, 2022, 38(4): 2074-2093

    2022

  76. [85]

    Bevplace++: Fast, robust, and lightweight lidar global localization for unmanned ground vehicles[J]

    Luo L, Cao S Y , Li X, et al. Bevplace++: Fast, robust, and lightweight lidar global localization for unmanned ground vehicles[J]. IEEE Trans- actions on Robotics, 2025

    2025

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.