Pith. sign in

REVIEW 3 major objections 4 minor 55 references

MISO: Multiresolution Submap Optimization for Efficient Globally Consistent Neural Implicit Reconstruction

T0 review · 3 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read MISO claims that fast global consistency in neural implicit SLAM comes from aligning and fusing multiresolution submaps directly in feature space, skipping geometry decoding for most of the optimization.

desk verdict MISO is a genuinely new neural implicit SLAM back-end with real speedups, but the feature-space alignment assumption is only partially tested and the baselines are not run under equal pre-training conditions. read the letter →

arxiv 2504.19104 v1 pith:K2X5UVYQ submitted 2025-04-27 cs.RO

Yulun Tian , Hanwen Cao , Sunghwan Kim , Nikolay Atanasov This is my paper · ORCID
classification cs.RO
keywords neuralimplicitSLAMmultiresolutionfeaturegridsubmapalignmentsigneddistancefunctionhierarchicaloptimizationfeature-spaceregistrationloopclosureposegraph
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

MISO claims that neural implicit SLAM can be made fast and globally consistent by optimizing the multiresolution feature grids themselves, rather than the geometry decoded from them. Locally, pre-trained hierarchical encoders initialize each level of the grid from residual inputs, so very few optimization epochs are needed. Globally, submaps are aligned by matching interpolated feature vectors in overlapping regions, with a short SDF-based refinement at the end. If the claim holds, large-scale neural signed-distance-function mapping becomes dramatically cheaper and stays accurate on real indoor and outdoor benchmarks.

What carries the argument

The central object is the multiresolution feature grid: a hierarchy of regular voxel grids whose trilinearly interpolated feature vectors are concatenated and decoded by a fixed MLP into signed distance values. Two mechanisms carry the argument. For local SLAM, a closed-form least-squares analysis shows the optimal initialization of a level's features is a linear function of coarser levels' residuals; this motivates learned 3D-CNN encoders that predict each level's grid from voxelized residuals. For global alignment, a correspondence-free cost compares the interpolated features of two submaps at overlapping grid vertices, and hierarchical coarse-to-fine optimization over submap poses minimizes it; only the final stage decodes SDF values, followed by feature-space averaging to fuse submaps in the world frame.

What would settle it

Take two overlapping submaps of the same scene built from disjoint depth or LiDAR observations with a known ground-truth relative pose, then optimize only the feature-based cost in (16) over that relative pose; if its global minimum does not sit at the ground-truth pose, or within its uncertainty, the comparability premise that carries MISO fails.

Watch

Extended reading notes

Core claim

The central claim is that both local submap optimization and global submap fusion can be carried out hierarchically in implicit feature space instead of in decoded geometry. For local SLAM, the paper proves that in a linear-least-squares special case, the optimal initialization of a level's features is a linear function of the residuals from coarser levels, and it converts this insight into learned hierarchical encoders that predict feature grids from voxelized residual inputs. For global alignment, two submaps are aligned by minimizing the squared Euclidean distance between trilinearly interpolated feature vectors at overlapping grid vertices, and only the final refinement stage uses predicted SDF values. The paper argues that this combination yields accuracy comparable to or better than SDF-based, correspondence-based, and ICP baselines while being substantially faster.

Load-bearing premise

The argument rests on the premise that feature vectors learned independently in different submaps are comparable, so aligning them in feature space is equivalent to aligning the underlying geometry; if that premise fails, the feature-based alignment cost can mislead the optimizer.

Editorial extensions

If this is right

  • Local submap optimization needs only about 20 epochs with learned hierarchical initialization, compared with 100 for the SDF and neural-point baselines, cutting back-end optimization time from tens of seconds to roughly one to eight seconds on ScanNet scenes.
  • Global submap alignment avoids decoding SDF for most iterations, bringing alignment time down from roughly 59–138 seconds to about 10–19 seconds while matching or improving final rotation and translation errors.
  • Feature-space fusion by weighted averaging of submap features yields a globally consistent SDF and mesh after loop closures, including on large indoor and outdoor LiDAR sequences.
  • The same hierarchical back-end is positioned to extend beyond SDF reconstruction to other scene properties such as radiance, semantics, or uncertainty.
  • In incremental mode, the same submap and alignment machinery improves trajectory accuracy substantially over the odometry initialization on outdoor Newer College sequences.

Reading between the lines

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

  • The paper does not say this, but if cross-submap feature comparability holds, MISO's feature-based alignment cost could let multiple agents fuse maps without ever exchanging raw geometry or decoded surfaces.
  • A direct extension of the linear-least-squares initialization result is that any grid-based implicit representation trained with a squared residual loss could be warmed up by a learned residual encoder, not just SDF grids.
  • The paper's own ablation notes mild domain mismatch after synthetic-only pretraining, which suggests alignment and initialization quality on a new deployment should track how well the pretraining distribution matches the scene.
  • Because alignment consumes only grid vertices, replacing dense grids with sparse or factored structures would likely preserve the speed gains while easing the memory cost the paper identifies as a limitation.
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

3 major / 4 minor

Summary. The paper proposes MISO, a hierarchical back-end for neural implicit SDF SLAM. Locally, each submap is a multiresolution feature grid, and features are initialized by learned hierarchical encoders before joint pose-feature optimization. Globally, submap base poses are aligned by minimizing feature-space costs at increasing resolutions (Eq. 16) followed by a final SDF-based refinement (Eq. 19), and submaps are fused by averaging features and decoding (Eqs. 20-21). The claims are that this avoids expensive geometric decoding for the bulk of the optimization, yielding substantial speedups and improved robustness over MIPS-Fusion, Vox-Fusion++, and ICP-style baselines, with evaluations on ScanNet, FastCaMo-Large, and Newer College.

Significance. If the central claims hold, MISO would be a practically valuable back-end for large-scale neural implicit SLAM: it directly exploits multiresolution features for alignment, reports large wall-clock reductions over geometry-based submap alignment, and includes a useful closed-form least-squares analysis (Proposition 1) that motivates the learned initialization. The manuscript is clearly written, uses public benchmarks, and provides ablations on the encoder/decoder contributions (Table V), the hierarchical alignment stages (Table VI), and the feature-space metric (Fig. 10). The main weaknesses are that the speed claims are confounded by pre-training and by per-scene decoder pre-training on Newer College, and that the key feature-comparability assumption for Eq. (16) is not established and is contradicted by the FastCaMo-Large coarse-level failure. These are evidential gaps rather than formal contradictions, so the claims remain plausible but not yet convincingly demonstrated.

major comments (3)
  1. [VI-C and Sec. V (Eq. 16)] The central speed advantage of global alignment rests on the feature-space cost in Eq. (16), but the manuscript states in Sec. VI-C that on FastCaMo-Large the coarse-level feature alignment is skipped 'as we observe it leads to degraded results on these datasets.' This is direct evidence that the feature cost can have spurious minima on the target domain, and Sec. V provides no mechanism or guarantee that independently optimized submap feature fields are comparable for the same surface. Please report feature-only alignment accuracy on FastCaMo-Large (without the SDF rescue stage in Algorithm 2 line 6), quantify the failure modes of Eq. (16), and either add an analysis of when feature comparability holds or restrict the robustness claim to settings where it is verified.
  2. [Table I and Table V] The local-mapping speed comparison in Table I is confounded by pre-training: MISO uses pre-trained encoder and decoder weights and is run for 20 epochs, while iSDF and Neural Points are trained from scratch for 100 epochs. Table V shows that the from-scratch variant No-ED reaches F-score 59.3 on scene 0011 at 100 epochs, essentially matching Full MISO at 10 epochs (59.4), and No-E with only a pre-trained decoder reaches 52.8 at 10 epochs. This suggests that a large part of the reported speed advantage is attributable to pre-training rather than to the hierarchical optimization scheme itself. Please add an ablation that fixes the pre-trained decoder across all variants and reports wall-clock time to reach a fixed F-score or loss, separating the contribution of hierarchical initialization from the contribution of pre-training.
  3. [VI-D and VI-A] The paper claims in Sec. VI-A that pre-trained weights are used on ScanNet and FastCaMo-Large 'without additional fine-tuning,' but Sec. VI-D states that on Newer College 'we first pre-train a decoder for each scene.' This per-scene decoder pre-training on the test domain weakens the stated generalization claim and makes the comparison with PIN-SLAM and KISS-ICP less direct, since those baselines do not receive per-scene pre-trained decoders. Please either rephrase the claims to acknowledge per-scene pre-training as a limitation, or add an experiment on Newer College without per-scene pre-training to show how much of the reported accuracy depends on this adaptation.
minor comments (4)
  1. [References] Reference [48] misspells the venue name as 'International Conference on Computer Cision'; this should be 'International Conference on Computer Vision' (or 'ICCV').
  2. [Sec. VI-E] The sentence 'We do not report the F-score for No-ED at epoch 10 because it fails to produce a mesh due to insufficient training' appears after Table VI but refers to the ablation in Table V; please move it to the discussion of Table V.
  3. [Sec. V, Eq. (16)] The definition of the vertex index set I^uv_l as 'level-l vertices in submap u that lie within the overlapping region' is vague; please specify how the overlapping region is computed from the current base poses, since this affects both the cost and its gradient.
  4. [Table IV] Table IV does not report optimization time, although the paper's headline claim is computational efficiency; please include runtime or convergence-time data for the Newer College experiments.

Circularity Check

1 steps flagged · score 2.0 of 10

No derivation-circularity in the core back-end; one evaluation-stage leak (per-scene decoder pre-training on Newer College) keeps the score low.

  1. fitted input called prediction [Section VI-D (Evaluation on Newer College Dataset), paragraph beginning 'For MISO, we first pre-train a decoder for each scene...']
    "For MISO, we first pre-train a decoder for each scene and then use the incremental implementation presented in Sec. IV-C to process each sequence."

    The Newer College benchmark is used to support the claim of scalability on large-scale real-world benchmarks, but before processing each sequence the decoder D_theta is pre-trained on that same sequence's observations. The decoder is therefore a fitted parameter carrying scene-specific SDF information, and the subsequent Chamfer-L1/F-score evaluation for that scene is not an out-of-sample prediction of the scene geometry. This is evaluation-level circularity: a parameter fitted to the test data is reused to produce the measured quantity. It does not collapse the core derivation, because submap features and poses are still optimized and the ScanNet and FastCaMo-Large experiments use Replica-only pre-training.

full rationale

The derivation chain is largely self-contained. Proposition 1 is a standard normal-equations solution of the linearized level-l subproblem, and the learned encoders are trained with the same residual objective (15), so the initialization is an amortized solver rather than a renamed target. The global alignment cost (16) relies on the empirical assumption that matching feature fields implies matching geometry, but the paper does not define alignment as feature alignment; it is validated against external ground-truth pose errors, and Sec. VI-C candidly reports that coarse feature alignment degrades results on FastCaMo-Large, which is evidence against, not evidence of, circularity. The only circularity-adjacent item is the per-scene decoder pre-training in Sec. VI-D: the decoder is fitted to each Newer College test sequence before that sequence is reconstructed and scored, so the Newer College results are partially contaminated by test-scene information. This is a real but localized evaluation leak, not a collapse of the algorithmic derivation; MISO's core speed and accuracy claims on ScanNet and FastCaMo-Large rest on Replica-only pre-training and remain independent.

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

The central claims rest on pre-trained components and on the comparability of independently optimized submap features. The listed free parameters are mostly standard defaults or experiment-specific settings; the trust-region radius is adjusted per experiment but not fitted to the outcome metrics. The feature-alignment assumption is the least supported premise.

free parameters (6)
  • trust region radius tau = set per experiment, e.g., from 1.0 deg/0.10 m to 17.0 deg/0.50 m
    Used in pose regularization (3) and adjusted based on the initial submap alignment error (Sec. VI-B).
  • SDF measurement weight w_sdf = 5.4
    Default from iSDF, used in the SDF supervision cost (4).
  • free-space bound penalty beta = 5.0
    Exponent in the lower-bound penalty (5).
  • grid resolutions = indoor 0.5 m and 0.1 m; outdoor 1.0 m and 0.2 m
    Spatial resolutions of the two feature grid levels (Sec. VI-A).
  • feature dimension d = 4
    Feature dimension at each grid level (Sec. VI-A).
  • alignment iteration counts kf,1, kf,2, ks = 45, 45, 10
    Iterations for coarse, fine, and SDF alignment stages in Algorithm 2 (Sec. VI-B).
assumptions (5)
  • standard math The solution to the linear least-squares problem (8) is given by the normal equations (Prop. 1).
    The closed-form (9) is the standard pseudo-inverse solution for linear least squares.
  • standard math SE(3) pose perturbations can be parameterized via the exponential map on the Lie algebra.
    Used to define local pose corrections epsilon in Problem 1 and Problem 2.
  • domain assumption For any pair of submaps to be well aligned, their implicit feature fields should also be aligned in the global frame.
    This is the key intuition behind the feature-based alignment cost (16); no proof is given that independently optimized submap features are comparable.
  • domain assumption A decoder and encoders pre-trained on synthetic Replica scenes transfer to real-world RGB-D and LiDAR data without fine-tuning.
    Local SLAM speed and accuracy rely on this transfer (Sec. IV-A and Sec. VI-A); the ablation in Table V shows performance can degrade at 100 epochs on scene 0011.
  • ad hoc to paper The linear-case closed-form initialization in Prop. 1 carries over to nonlinear decoders when the linear map is replaced by a trained neural network.
    Section IV-B motivates the learned encoder E_phi as an approximation of the linear solution, but no guarantee or bound is provided for the nonlinear case.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MISO: Multiresolution Submap Optimization for Efficient Globally Consistent Neural Implicit Reconstruction." pith.science (2026). https://pith.science/paper/K2X5UVYQ

@misc{pith2026250419104,
  author       = {Pith},
  title        = {Pith review of: MISO: Multiresolution Submap Optimization for Efficient Globally Consistent Neural Implicit Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K2X5UVYQ}},
  note         = {Machine review of arXiv:2504.19104}
}
read the original abstract

Neural implicit representations have had a significant impact on simultaneous localization and mapping (SLAM) by enabling robots to build continuous, differentiable, and high-fidelity 3D maps from sensor data. However, as the scale and complexity of the environment increase, neural SLAM approaches face renewed challenges in the back-end optimization process to keep up with runtime requirements and maintain global consistency. We introduce MISO, a hierarchical optimization approach that leverages multiresolution submaps to achieve efficient and scalable neural implicit reconstruction. For local SLAM within each submap, we develop a hierarchical optimization scheme with learned initialization that substantially reduces the time needed to optimize the implicit submap features. To correct estimation drift globally, we develop a hierarchical method to align and fuse the multiresolution submaps, leading to substantial acceleration by avoiding the need to decode the full scene geometry. MISO significantly improves computational efficiency and estimation accuracy of neural signed distance function (SDF) SLAM on large-scale real-world benchmarks.

Figures

Figures reproduced from arXiv: 2504.19104 by the authors.

Figure 1
Figure 1. Demonstration of MISO on the FastCaMo-Large dataset [ [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of MISO. (a) Given point cloud observations, MISO performs local hierarchical SLAM within a submap represented as a multiresolution feature grid (Sec. IV). (b) Given locally optimized submaps, MISO performs global alignment and fusion across submaps to eliminate estimation drift and achieve globally consistent scene reconstruction (Sec. V). seek to jointly refine the robot’s pose estimates and the submap fe… view at source ↗
Figure 3
Figure 3. Illustration of the level- [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Visualization of estimated SDF at a fixed height on ScanNet [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 6
Figure 6. Figure 6: Evaluation of submap alignment on ScanNet scene 0011 under [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: 3D mesh reconstruction with oriented submap bounding [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 9
Figure 9. Figure 9: Qualitative evaluation on the Newer College dataset [ [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 10
Figure 10. Figure 10: Ablation study on hierarchical alignment on ScanNet [ [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: Qualitative comparison on additional ScanNet scenes 0000, 0011, and 0024, each shown in a row. For each method, a horizontal [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 51 canonical work pages

  1. [1]

    MIPS-Fusion: Multi-implicit-submaps for scalable and robust online neural RGB-D reconstruction,

    Y . Tang, J. Zhang, Z. Yu, H. Wang, and K. Xu, “MIPS-Fusion: Multi-implicit-submaps for scalable and robust online neural RGB-D reconstruction,” ACM Transactions on Graphics (TOG) , 2023. 1, 2, 3, 6, 7, 8, 9, 10, 14

    work page 2023

  2. [2]

    Neural fields in visual computing and beyond,

    Y . Xie, T. Takikawa, S. Saito, O. Litany, S. Yan, N. Khan, F. Tombari, J. Tompkin, V . Sitzmann, and S. Sridhar, “Neural fields in visual computing and beyond,” Computer Graphics Forum (CGF) , 2022. 1, 2

    work page 2022

  3. [3]

    How NeRFs and 3D Gaussian Splatting are reshaping SLAM: A survey,

    F. Tosi, Y . Zhang, Z. Gong, E. Sandstr ¨om, S. Mattoccia, M. R. Oswald, and M. Poggi, “How NeRFs and 3D Gaussian Splatting are reshaping SLAM: A survey,” arXiv preprint arXiv:2402.13255 , 2024. 1, 2, 3

    arXiv 2024

  4. [4]

    Hierarchical SLAM: Real- time accurate mapping of large environments,

    C. Estrada, J. Neira, and J. D. Tard ´os, “Hierarchical SLAM: Real- time accurate mapping of large environments,” IEEE Transactions on Robotics (T-RO), 2005. 1

    work page 2005

  5. [5]

    A multilevel relaxation algorithm for simultaneous localization and mapping,

    U. Frese, P. Larsson, and T. Duckett, “A multilevel relaxation algorithm for simultaneous localization and mapping,” IEEE Transactions on Robotics (T-RO), 2005. 1

    work page 2005

  6. [6]

    Hierarchical optimization on manifolds for online 2D and 3D mapping,

    G. Grisetti, R. K ¨ummerle, C. Stachniss, U. Frese, and C. Hertzberg, “Hierarchical optimization on manifolds for online 2D and 3D mapping,” in International Conference on Robotics and Automation (ICRA) , 2010. 1

    work page 2010

  7. [7]

    Neural geometric level of detail: Real-time rendering with implicit 3D shapes,

    T. Takikawa, J. Litalien, K. Yin, K. Kreis, C. Loop, D. Nowrouzezahrai, A. Jacobson, M. McGuire, and S. Fidler, “Neural geometric level of detail: Real-time rendering with implicit 3D shapes,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021. 1, 2, 3

    work page 2021

  8. [8]

    Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction,

    C. Sun, M. Sun, and H.-T. Chen, “Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 1, 2, 3

    work page 2022

Show all 55 references
  1. [9]

    Instant neural graphics primitives with a multiresolution hash encoding,

    T. M ¨uller, A. Evans, C. Schied, and A. Keller, “Instant neural graphics primitives with a multiresolution hash encoding,” ACM Transactions on Graphics (TOG), 2022. 1, 2, 3

    2022

  2. [10]

    V ox-Fusion++: V oxel-based neural implicit dense tracking and mapping with multi-maps,

    H. Zhai, H. Li, X. Yang, G. Huang, Y . Ming, H. Bao, and G. Zhang, “V ox-Fusion++: V oxel-based neural implicit dense tracking and mapping with multi-maps,” arXiv preprint arXiv:2403.12536, 2024. 1, 2, 3, 6, 7, 9

    2024 arXiv

  3. [11]

    iSDF: Real-time neural signed distance fields for robot perception,

    J. Ortiz, A. Clegg, J. Dong, E. Sucar, D. Novotny, M. Zollhoefer, and M. Mukadam, “iSDF: Real-time neural signed distance fields for robot perception,” in Robotics: Science and Systems (RSS) , 2022. 1, 2, 4, 7, 8, 15

    2022

  4. [12]

    3D Gaussian splatting for real-time radiance field rendering,

    B. Kerbl, G. Kopanas, T. Leimk ¨uhler, and G. Drettakis, “3D Gaussian splatting for real-time radiance field rendering,” ACM Transactions on Graphics (TOG), 2023. 2

    2023

  5. [13]

    DeepSDF: Learning continuous signed distance functions for shape representation,

    J. J. Park, P. Florence, J. Straub, R. Newcombe, and S. Lovegrove, “DeepSDF: Learning continuous signed distance functions for shape representation,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019. 2

    2019

  6. [14]

    Occupancy networks: Learning 3D reconstruction in function space,

    L. Mescheder, M. Oechsle, M. Niemeyer, S. Nowozin, and A. Geiger, “Occupancy networks: Learning 3D reconstruction in function space,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019. 2

    2019

  7. [15]

    NeRF: Representing scenes as neural radiance fields for view synthesis,

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, and R. Ng, “NeRF: Representing scenes as neural radiance fields for view synthesis,” Communications of the ACM (CACM) , 2021. 2

    2021

  8. [16]

    Neural RGB-D surface reconstruction,

    D. Azinovi ´c, R. Martin-Brualla, D. B. Goldman, M. Nießner, and J. Thies, “Neural RGB-D surface reconstruction,” in IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR) , 2022. 2

    2022

  9. [17]

    Plenoxels: Radiance fields without neural networks,

    S. Fridovich-Keil, A. Yu, M. Tancik, Q. Chen, B. Recht, and A. Kanazawa, “Plenoxels: Radiance fields without neural networks,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 2

    2022

  10. [18]

    K-Planes: Explicit radiance fields in space, time, and appearance,

    S. Fridovich-Keil, G. Meanti, F. R. Warburg, B. Recht, and A. Kanazawa, “K-Planes: Explicit radiance fields in space, time, and appearance,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2

    2023

  11. [19]

    TensoRF: Tensorial radiance fields,

    A. Chen, Z. Xu, A. Geiger, J. Yu, and H. Su, “TensoRF: Tensorial radiance fields,” in European Conference on Computer Vision (ECCV) ,

  12. [20]

    Point-NeRF: Point-based neural radiance fields,

    Q. Xu, Z. Xu, J. Philip, S. Bi, Z. Shu, K. Sunkavalli, and U. Neu- mann, “Point-NeRF: Point-based neural radiance fields,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 2

    2022

  13. [21]

    Neuralangelo: High-fidelity neural surface reconstruction,

    Z. Li, T. M ¨uller, A. Evans, R. H. Taylor, M. Unberath, M.-Y . Liu, and C.-H. Lin, “Neuralangelo: High-fidelity neural surface reconstruction,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2, 4

    2023

  14. [22]

    Plenoc- trees for real-time rendering of neural radiance fields,

    A. Yu, R. Li, M. Tancik, H. Li, R. Ng, and A. Kanazawa, “Plenoc- trees for real-time rendering of neural radiance fields,” in IEEE/CVF International Conference on Computer Vision (ICCV) , 2021. 2

    2021

  15. [23]

    H2-mapping: Real-time dense mapping using hierarchical hybrid rep- resentation,

    C. Jiang, H. Zhang, P. Liu, Z. Yu, H. Cheng, B. Zhou, and S. Shen, “H2-mapping: Real-time dense mapping using hierarchical hybrid rep- resentation,” Robotics and Automation Letters , 2023. 2

    2023

  16. [24]

    GO-Surf: Neural feature grid optimization for fast, high-fidelity RGB-D surface reconstruction,

    J. Wang, T. Bleja, and L. Agapito, “GO-Surf: Neural feature grid optimization for fast, high-fidelity RGB-D surface reconstruction,” in IEEE International Conference on 3D Vision (3DV) , 2022. 2

    2022

  17. [25]

    iMAP: Implicit mapping and positioning in real-time,

    E. Sucar, S. Liu, J. Ortiz, and A. J. Davison, “iMAP: Implicit mapping and positioning in real-time,” in IEEE/CVF International Conference on Computer Vision (ICCV) , 2021. 2

    2021

  18. [26]

    V ox-Fusion: Dense tracking and mapping with voxel-based neural implicit repre- sentation,

    X. Yang, H. Li, H. Zhai, Y . Ming, Y . Liu, and G. Zhang, “V ox-Fusion: Dense tracking and mapping with voxel-based neural implicit repre- sentation,” in IEEE International Symposium on Mixed and Augmented Reality (ISMAR), 2022. 2

    2022

  19. [27]

    NICER-SLAM: Neural implicit scene encoding for RGB SLAM,

    Z. Zhu, S. Peng, V . Larsson, Z. Cui, M. R. Oswald, A. Geiger, and M. Pollefeys, “NICER-SLAM: Neural implicit scene encoding for RGB SLAM,” in International Conference on 3D Vision (3DV) , 2024. 2

    2024

  20. [28]

    NeRF-LOAM: Neural implicit representation for large-scale incremental lidar odometry and mapping,

    J. Deng, Q. Wu, X. Chen, S. Xia, Z. Sun, G. Liu, W. Yu, and L. Pei, “NeRF-LOAM: Neural implicit representation for large-scale incremental lidar odometry and mapping,” in IEEE/CVF International Conference on Computer Vision (ICCV) , 2023. 2

    2023

  21. [29]

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

    Y . Pan, X. Zhong, L. Wiesmann, T. Posewsky, J. Behley, and C. Stach- niss, “PIN-SLAM: LiDAR SLAM using a point-based implicit neural representation for achieving global map consistency,”IEEE Transactions on Robotics (T-RO), 2024. 2, 3, 6, 7, 8, 10, 15

    2024

  22. [30]

    PINGS: Gaussian Splatting Meets Distance Fields within a Point-Based Implicit Neural Map,

    Y . Pan, X. Zhong, L. Jin, L. Wiesmann, M. Popovi ´c, J. Behley, and C. Stachniss, “PINGS: Gaussian Splatting Meets Distance Fields within a Point-Based Implicit Neural Map,” arXiv preprint arXiv:2502.05752 ,

    arXiv

  23. [31]

    ESLAM: Efficient dense SLAM system based on hybrid representation of signed distance fields,

    M. M. Johari, C. Carta, and F. Fleuret, “ESLAM: Efficient dense SLAM system based on hybrid representation of signed distance fields,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2

    2023

  24. [32]

    Co-SLAM: Joint coordinate and sparse parametric encodings for neural real-time SLAM,

    H. Wang, J. Wang, and L. Agapito, “Co-SLAM: Joint coordinate and sparse parametric encodings for neural real-time SLAM,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2

    2023

  25. [33]

    GO-SLAM: Global optimization for consistent 3D instant reconstruction,

    Y . Zhang, F. Tosi, S. Mattoccia, and M. Poggi, “GO-SLAM: Global optimization for consistent 3D instant reconstruction,” in IEEE/CVF International Conference on Computer Vision (ICCV) , 2023. 2

    2023

  26. [34]

    Real-time large-scale dense 3D reconstruction with loop closure,

    O. K ¨ahler, V . A. Prisacariu, and D. W. Murray, “Real-time large-scale dense 3D reconstruction with loop closure,” in European Conference on Computer Vision (ECCV) , 2016. 3

    2016

  27. [35]

    Loopy-SLAM: Dense neural SLAM with loop closures,

    L. Liso, E. Sandstr ¨om, V . Yugay, L. Van Gool, and M. R. Oswald, “Loopy-SLAM: Dense neural SLAM with loop closures,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 3

    2024

  28. [36]

    PLGSLAM: Progressive neural scene represenation with local to global bundle adjustment,

    T. Deng, G. Shen, T. Qin, J. Wang, W. Zhao, J. Wang, D. Wang, and W. Chen, “PLGSLAM: Progressive neural scene represenation with local to global bundle adjustment,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2024. 3

    2024

  29. [37]

    NEWTON: Neural view-centric mapping for on-the-fly large-scale SLAM,

    H. Matsuki, K. Tateno, M. Niemeyer, and F. Tombari, “NEWTON: Neural view-centric mapping for on-the-fly large-scale SLAM,” IEEE Robotics and Automation Letters (RA-L) , 2024. 3

    2024

  30. [38]

    Efficient map fusion for multiple implicit slam agents,

    S. Liu and J. Zhu, “Efficient map fusion for multiple implicit slam agents,” Transactions on Intelligent Vehicles (T-IV) , 2023. 3

    2023

  31. [39]

    CP-SLAM: Collaborative neural point-based SLAM system,

    J. Hu, M. Mao, H. Bao, G. Zhang, and Z. Cui, “CP-SLAM: Collaborative neural point-based SLAM system,” in Advances in Neural Information Processing Systems (NeurIPS) , 2024. 3

    2024

  32. [40]

    NICE-SLAM: Neural implicit scalable encoding for SLAM,

    Z. Zhu, S. Peng, V . Larsson, W. Xu, H. Bao, Z. Cui, M. R. Oswald, and M. Pollefeys, “NICE-SLAM: Neural implicit scalable encoding for SLAM,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 3

    2022

  33. [41]

    Nocedal and S

    J. Nocedal and S. J. Wright, Numerical optimization. Springer, 1999. 4

    1999

  34. [42]

    Adam: A method for stochastic optimization,

    D. P. Kingma, “Adam: A method for stochastic optimization,” in International Conference on Learning Representations (ICLR) , 2014. 4, 13

    2014

  35. [43]

    PyTorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, et al. , “PyTorch: An imperative style, high-performance deep learning library,” in Advances in Neural Information Processing Systems (NeurIPS) , 2019. 4, 6, 8

    2019

  36. [44]

    Implicit ge- ometric regularization for learning shapes,

    A. Gropp, L. Yariv, N. Haim, M. Atzmon, and Y . Lipman, “Implicit ge- ometric regularization for learning shapes,” in International Conference on Machine Learning (ICML) , 2020. 4

    2020

  37. [45]

    Con- volutional occupancy networks,

    S. Peng, M. Niemeyer, L. Mescheder, M. Pollefeys, and A. Geiger, “Con- volutional occupancy networks,” in European Conference on Computer Vision (ECCV), 2020. 5

    2020

  38. [46]

    NV AE: A deep hierarchical variational autoencoder,

    A. Vahdat and J. Kautz, “NV AE: A deep hierarchical variational autoencoder,” in Advances in Neural Information Processing Systems (NeurIPS), 2020. 5

    2020

  39. [47]

    ScanNet: Richly-annotated 3D reconstructions of indoor scenes,

    A. Dai, A. X. Chang, M. Savva, M. Halber, T. Funkhouser, and M. Nießner, “ScanNet: Richly-annotated 3D reconstructions of indoor scenes,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2017. 7, 8, 9, 11, 12, 14

    2017

  40. [48]

    Colored point cloud registration revisited,

    J. Park, Q.-Y . Zhou, and V . Koltun, “Colored point cloud registration revisited,” in International Conference on Computer Cision , 2017. 8, 9

    2017

  41. [49]

    The Replica dataset: A digital replica of indoor spaces,

    J. Straub, T. Whelan, L. Ma, Y . Chen, E. Wijmans, S. Green, J. J. Engel, R. Mur-Artal, C. Ren, S. Verma, A. Clarkson, M. Yan, B. Budge, Y . Yan, X. Pan, J. Yon, Y . Zou, K. Leon, N. Carter, J. Briales, T. Gillingham, E. Mueggler, L. Pesqueira, M. Savva, D. Batra, H. M. Strasd...

    1906 arXiv

  42. [50]

    Multi-camera li- dar inertial extension to the newer college dataset,

    L. Zhang, M. Camurri, D. Wisth, and M. Fallon, “Multi-camera li- dar inertial extension to the newer college dataset,” arXiv preprint arXiv:2112.08854, 2021. 7, 10

    2021 arXiv

  43. [51]

    Robust reconstruction of indoor scenes,

    S. Choi, Q.-Y . Zhou, and V . Koltun, “Robust reconstruction of indoor scenes,” in Conference on computer vision and pattern recognition ,

  44. [52]

    Open3D: A modern library for 3D data processing,

    Q.-Y . Zhou, J. Park, and V . Koltun, “Open3D: A modern library for 3D data processing,” arXiv preprint arXiv:1801.09847 , 2018. 7, 10

    2018 arXiv

  45. [53]

    KISS-ICP: In defense of point-to-point ICP–simple, accurate, and robust registration if done the right way,

    I. Vizzo, T. Guadagnino, B. Mersch, L. Wiesmann, J. Behley, and C. Stachniss, “KISS-ICP: In defense of point-to-point ICP–simple, accurate, and robust registration if done the right way,” IEEE Robotics and Automation Letters (RA-L) , 2023. 10 APPENDIX A IMPLEMENTATION AND TRAI...

    2023

  46. [54]

    During training, we employ a coarse-to-fine strategy where all fine level features are activated after 200 epochs

    with a learning rate of 10−3 and train for a total of 1200 epochs. During training, we employ a coarse-to-fine strategy where all fine level features are activated after 200 epochs. After training completes, the grid features Fs are discarded and only decoder parameters θ is s...

  47. [55]

    APPENDIX C ADDITIONAL RESULTS In Fig

    (27) Thus, the solution is obtained by solving its normal equations, yielding the final expression in (9). APPENDIX C ADDITIONAL RESULTS In Fig. 11, we show qualitative SDF visualizations on additional scenes from ScanNet [47] under noisy poses. For each method, a horizontal s...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.