Pith. sign in

REVIEW 3 major objections 6 minor 44 references

FrontierNet: Learning Visual Cues to Explore

T0 review · 3 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read FrontierNet claims that exploration goals can be read directly from posed camera images, with a learned model marking frontier pixels and predicting how much unknown space each one would reveal, replacing dense 3D frontier extraction.

desk verdict A solid, incremental-but-real step: learning to propose frontiers and predict info gain directly from images, with experiments that mostly back the headline. read the letter →

arxiv 2501.04597 v2 pith:5ZUVQL3C submitted 2025-01-08 cs.RO cs.CV

classification cs.ROcs.CV
keywords autonomousexplorationfrontierdetectioninformationgainpredictionmonoculardepthestimationvisual-onlyimage-to-imagenext-best-viewplanningmap-free
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

The paper argues that frontier-based exploration need not start from a dense 3D map. It proposes FrontierNet, a learned image-to-image model that marks frontier pixels in an RGB frame, enriched with a monocular depth prior, and assigns each one an information gain: the amount of unknown volume that a viewpoint at that frontier would reveal. If the claim is right, robots can decide where to go from visual appearance alone, and they should map more efficiently because they prioritize high-yield frontiers. The authors report roughly 15 percentage points higher Vox@50 than the best baseline across ten held-out scenes, and a map-free variant that still explores without any dense occupancy grid.

What carries the argument

The central object is a two-head image-to-image network with a shared encoder-decoder backbone: one head regresses a truncated distance field whose zero level set marks frontier pixels, and the other classifies per-pixel information gain into 11 discrete bins. Both heads are supervised by ground truth generated from voxelized 3D scenes, where frontier voxels are projected to pixels and refined by a depth-discontinuity mask. At inference, the depth gradient supplies the 2D viewing direction for each frontier pixel, and averaging foreground and background depths lifts clustered 2D frontiers into sparse 3D goal poses, so the image-space proposal becomes a concrete exploration goal without dense map sampling.

What would settle it

Run all methods on the same ten scenes with an identical low-level planner, identical frontier update rules, and identical occupancy-map quality, giving every method the same depth input; if FrontierNet's Vox@50 lead over the best baseline falls well short of the claimed ~15 percentage points under that common protocol, the central efficiency claim is not supported.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the 2D projection of 3D frontier voxels is learnable directly on the image plane, together with a ranking of how much new volume each frontier would expose. FrontierNet treats frontier detection as a distance-field regression from an RGB-plus-depth image and treats information gain as a binned classification problem, then anchors the resulting 2D frontier clusters into sparse 3D goal poses by reading foreground and background depths and gradient directions. The same visual-only pipeline, freed from dense map operations, reports the highest Vox@25, Vox@50, and success rate among the compared methods, and keeps most of its advantage when the depth input is a monocular prediction rather than simulator depth. The paper additionally claims the system explores reasonably in a map-free configuration, keeping only a sparse frontier tree and trajectory memory.

Load-bearing premise

The training labels only keep frontier pixels that coincide with a depth discontinuity ($F = F_p \cap F_d$), so a useful frontier with no depth gradient, such as a wide open doorway into a featureless room at similar range, is removed from the ground truth and can never be proposed by the model.

Editorial extensions

If this is right

  • If the central claim is correct, exploration goal extraction no longer requires dense 3D map operations, so systems can operate with predicted monocular depth and reduced computational or storage overhead.
  • Prioritizing frontiers by predicted information gain yields early exploration gains, with about 15 percentage points higher Vox@50 than the best baselines across ten unseen HM3D scenes.
  • A map-free variant that keeps only a sparse frontier tree and trajectory memory still explores, opening the approach to settings without a dense map or under severe map degradation.
  • Because the input is individual posed images, the same pipeline could be retrained or fine-tuned for other image-based goals such as object search or goal-directed navigation.
  • The information-gain ranking is learned from visual appearance, so texture, color, and semantic context can inform how much unknown space a frontier hides, something 3D map metrics rarely capture.

Reading between the lines

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

  • An extension the paper leaves implicit: because the training labels only keep frontiers that coincide with a depth discontinuity ($F = F_p \cap F_d$), the model would likely miss an open doorway into a featureless room at similar range; relaxing that intersection with semantic boundary labels would test how much coverage is lost.
  • Because the information-gain head learns from RGB appearance, the same architecture should transfer to object-seeking or semantic exploration by swapping the gain label definition, without changing the proposal mechanism.
  • The reported 15-point margin compares FrontierNet, in one variant using predicted depth, against baselines given simulator depth; a strictly matched protocol where every method receives identical map quality would separate the learned-frontier effect from planning-infrastructure effects.
  • I read the map-free result as evidence that exploration can be treated as a sparse image-anchored graph problem, which may combine naturally with topological SLAM or vision-language navigation, though the paper does not test those combinations.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. FrontierNet proposes a learning-based exploration system that replaces dense 3D map operations for goal extraction with a 2D image-based frontier proposer and information-gain predictor. Given a posed RGB image augmented by a monocular depth prior, a two-head UNet predicts a distance field to frontier pixels and a discretized information-gain map. The predictions are clustered in 2D, lifted to sparse 3D viewpoints using depth gradients and depths, and then managed by a frontier-update and utility-based planner. The system is evaluated on 10 HM3D scenes against Classic, NBVP, SEER, and a SEER re-implementation under identical planning conditions, using both simulator depth and monocular depth, plus a map-free variant and a real Spot robot deployment. The headline result is roughly 15 percentage points higher Vox@50 mapped volume than the best baseline.

Significance. If validated, the paper offers a refreshing alternative to classical 3D-map-centric exploration: sparse 2D visual predictions, combined with monocular depth, can produce competitive or better early-stage exploration efficiency. The evaluation is more extensive than typical for this area (10 scenes vs. the usual 1–2), includes a controlled re-implementation of SEER with the same planner, performs ablations on input modalities and system components, and provides a real-robot demonstration plus public code. The main scientific risks are the depth-discontinuity training filter, which may restrict the model to a subset of visually identifiable frontiers, and the unspecified handling of never-reaching trials in the Vox@k metric. Both are addressable with additional analysis and precise protocol specification. The reader's concern about F = Fp ∩ Fd is real: the paper's own text acknowledges the 'typically' depth-gap assumption but does not quantify how much of the frontier space is removed by the filter.

major comments (3)
  1. [Section III-D, Fig. 4] The refined frontier mask is defined as F = Fp ∩ Fd, with Fd a thresholded depth-gradient mask. Because the distance-field head and info-gain head are supervised only from this intersection, the learned proposer can never output a frontier pixel whose projected location lacks a depth discontinuity. The paper justifies this by saying frontiers are 'typically' associated with depth gaps, but it reports no coverage statistic: what fraction of Fp frontier pixels survive the intersection, and are high-info-gain frontier voxels preferentially removed? This is load-bearing for the central claim of learning frontier proposal from visual cues: in an open doorway into a flat, untextured region at similar depth, or under sensor degradation, a meaningful frontier can exist without a depth edge, and the model is structurally unable to propose it. Please quantify the filter's effect on the training distribution and either relax the filter or explicitly restrict the claim to depth-discontinuity frontiers.
  2. [Section IV-A, Vox@k definition] Vox@k is defined using a per-scene 'average step count across all methods at which 25% and 50% volume coverage is reached.' The manuscript does not specify what happens to trials that never reach the target coverage: are they excluded from the average, assigned the step limit, counted as infinity? Similarly, for a method that never reaches coverage by the averaged threshold, is its Vox@k value capped at the achieved volume at the maximum step count, or is it set to zero? Because the headline 15% advantage at Vox@50 depends on this protocol, the ambiguity could materially change the comparison. Please state the exact handling of non-reaching trials and, if feasible, provide a sensitivity analysis with alternative threshold definitions.
  3. [Section III-D, info-gain ground truth] The info-gain label is approximated by sub-sampling 10% of the frontier voxels, computing their gains, and linearly interpolating to the remainder. The paper justifies this by the assumption that spatially close frontier voxels have similar gains. This assumption is plausible in open regions but can fail at complex boundaries where adjacent voxels face disjoint unknown regions. Since the learned info-gain head is directly supervised by these labels, interpolation errors can systematically bias the ranking that drives the planner. Please report a sensitivity analysis of the subsampling ratio (e.g., 5% vs. 20%) or, at minimum, provide label-error statistics on the validation set.
minor comments (6)
  1. [Section III-C] The symbol D is used both for the distance field (Eq. 1) and for depth in later sections (e.g., d_b, d_f in Section III-E3). Although the depth symbols are lowercase, the proximity in notation may confuse readers; consider using a distinct symbol for the distance field, such as DF.
  2. [Section IV-A] The scene-specific step limit is mentioned but never defined. Please report the step limits used for each of the 10 scenes, as they directly affect the Vox@100 and success-rate numbers.
  3. [Table I] In the SEER row for scene 879, the entry '×' is not explained in the caption or text. Please clarify whether this denotes a planning failure, a failure to produce a map, or a missing trial.
  4. [Section IV-C] The real-world validation is qualitative; please consider reporting quantitative metrics from the Spot deployment, such as mapped volume over time, total path length, or number of frontiers visited, to support the claim of sim-to-real transfer.
  5. [Section IV-A] The paper states that 'several initial poses' are used per scene but does not specify their number or locations. Listing these initial poses is important for reproducibility.
  6. [Eq. (2)] The pixel set S used in Eq. (2) is not introduced before its use; please define it explicitly (e.g., as the set of pixels in the binary frontier mask F).

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the pipeline trains on 3D-derived labels and is evaluated against external baselines under a shared planner.

full rationale

Walking the paper's derivation chain: FrontierNet is trained on ground-truth frontier masks and info-gain maps generated by projecting 3D frontier voxels and ray-casting privileged occupancy (Section III-D), and the exploration planner consumes the model's predicted distance field and info-gain classification (Sections III-E and III-F). The headline Vox@k results compare the resulting exploration behavior against Classic, NBVP, SEER, and a SEER-frontier variant under the same planner, frontier-update logic, and test protocol (Section IV-B), so the evaluation is not a re-identification of the training labels. No equation in the paper defines the predicted distance field or info gain as the evaluation metric, and no parameter fitted to the benchmark scenes is renamed as a prediction. The monocular depth prior from Metric3D v2 is an external, benchmarked model; the co-authorship overlap with H. Chen does not make that citation load-bearing, and the paper also reports results with perfect depth. The F = Fp ∩ Fd refinement in Section III-D is a generalization assumption about frontiers coinciding with depth discontinuities, not a circular reduction, because it constrains the training label distribution rather than equating any predicted quantity with the measured output. Therefore no specific circular step can be exhibited.

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

The system contributes a trained model; the listed hyperparameters and data-generation approximations are hand-chosen and not sensitivity-tested. No new physical entities are introduced. The main axioms concern label generation and the evaluation metric, not new physics.

free parameters (6)
  • alpha (loss weight) = not specified
    Balances distance-field L1 loss and info-gain cross-entropy plus Dice loss in Eq. 5; no sensitivity study is given.
  • K (info gain bins) = 11
    Discretizes info gain into 11 classes in Eq. 4; no ablation on bin count is provided.
  • l (frontier inclusion threshold) = not specified
    Threshold on predicted distance field to recover frontier pixels in Eq. 6; controls precision and recall.
  • r (ray-casting range) = not specified
    Retention radius for frontier voxels near each ray during ground-truth generation in Section III-D; controls extent of the info gain map.
  • merge thresholds and g_min = not specified
    Distance and angle thresholds for merging 3D frontiers and minimum info gain for validity in Section III-F; affect frontier list dynamics.
  • depth gradient threshold for F_d = not specified
    Thresholding the depth gradient creates the discontinuity mask; the exact threshold is not given.
assumptions (5)
  • domain assumption Every 2D frontier pixel is the projection of a 3D frontier voxel, and all informative frontiers appear as depth discontinuities (F = Fp ∩ Fd).
    Section III-D, Fig. 4. The learned frontier mask is trained only on the intersection of projected frontier voxels and the depth discontinuity mask.
  • domain assumption The optimal viewing direction at a frontier is toward unknown space, given by the inverse of the local depth gradient.
    Section III-E1. Used to generate viewpoints without 3D sampling.
  • domain assumption Spatially close frontier voxels have similar info gain, so a 10% subsample plus linear interpolation is a valid label.
    Section III-D. This approximation defines the training target for the info-gain head.
  • domain assumption Mapped volume (|V_known|) is the correct measure of exploration success.
    Section III-A. Follows prior works [11], [14], [15]; the entire evaluation rests on this proxy.
  • domain assumption Monocular depth priors are accurate enough for 3D lifting and path planning.
    Section III-E3 acknowledges scale inaccuracies but claims robustness via averaging; default experiments use Metric3D v2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FrontierNet: Learning Visual Cues to Explore." pith.science (2026). https://pith.science/paper/5ZUVQL3C

@misc{pith2026250104597,
  author       = {Pith},
  title        = {Pith review of: FrontierNet: Learning Visual Cues to Explore},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5ZUVQL3C}},
  note         = {Machine review of arXiv:2501.04597}
}
read the original abstract

Exploration of unknown environments is crucial for autonomous robots; it allows them to actively reason and decide on what new data to acquire for different tasks, such as mapping, object discovery, and environmental assessment. Existing solutions, such as frontier-based exploration approaches, rely heavily on 3D map operations, which are limited by map quality and, more critically, often overlook valuable context from visual cues. This work aims at leveraging 2D visual cues for efficient autonomous exploration, addressing the limitations of extracting goal poses from a 3D map. We propose a visual-only frontier-based exploration system, with FrontierNet as its core component. FrontierNet is a learning-based model that (i) proposes frontiers, and (ii) predicts their information gain, from posed RGB images enhanced by monocular depth priors. Our approach provides an alternative to existing 3D-dependent goal-extraction approaches, achieving a 15\% improvement in early-stage exploration efficiency, as validated through extensive simulations and real-world experiments. The project is available at https://github.com/cvg/FrontierNet.

Figures

Figures reproduced from arXiv: 2501.04597 by the authors.

Figure 1
Figure 1. Top: FrontierNet processes a RGB image (left) to propose frontier pixels and their information gain (middle), registering candidate goal viewpoints with varying priorities in 3D (right). Bottom: Using FrontierNet, our exploration system prioritizes visiting unknown regions with greater potential of unmapped volume, achieving higher efficiency. poses or paths within the map and select the most suitable ones [9]. Thes… view at source ↗
Figure 2
Figure 2. FrontierNet learns to propose regions for exploration from visual cues in RGB [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 4
Figure 4. Ground Truth Generation. For a sampled camera pose in the voxelized scene, 3D frontier voxels are calculated and projected onto the camera frame using ground truth 3D occupancy grid. Merging the projection with the depth discontinuity mask produces a refined and less noisy frontier pixels mask 𝐅, which is used to calculate the distance field map 𝐃. Additionally, projecting the info gain of each frontier voxel onto t… view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: 3D Frontier Generation. Each frontier pixel is assigned a 2D viewing angle derived from the depth gradient. Combined with the info gain, 2D clustering is applied to obtain sparse 2D frontier clusters with associated viewing directions (middle). The foreground and backg…
Figure 7
Figure 7. Figure 7: Path Planning. When the robot is at pose 𝐱𝑟 , the next goal frontier 𝐟𝑘 , proposed and registered by its previous pose 𝐱𝑡 , lies outside the current 3D occupancy map (red voxelgrid). The planner samples points (black dots) backward along the edge (𝐱𝑡 , 𝐟𝑘 ) until it fi…
Figure 10
Figure 10. Figure 10: Map-Free Exploration Example. Examples across six different scenes (Top row: scenes 804, 827, 879; bottom row: scenes 883, 880, 876.) No dense 3D map is maintained during exploration; the reconstructions shown serve only as visualizations. 804 827 876 879 880 883 Mean…
Figure 9
Figure 9. Figure 9: Performance comparison of different configurations on scene 824 (left) and scene [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 11
Figure 11. Figure 11: Real-world Validation Result. Exploration process of a quadrupedal robot in a real-world environment. Top: Floor plan. Bottom: Reconstructed map and exploration path from TSDF integration using monocular depth prediction. Colored boxes indicate key correspondences bet…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 28 canonical work pages

  1. [1]

    Semantic belief behavior graph: Enabling au- tonomous robot inspection in unknown environments,

    M. F. Ginting, D. D. Fan, S.-K. Kim, M. J. Kochenderfer, and A.- a. Agha-mohammadi, “Semantic belief behavior graph: Enabling au- tonomous robot inspection in unknown environments,”arXiv preprint arXiv:2401.17191, 2024

  2. [2]

    Infrastructure robotics: Research challenges and opportunities,

    D. Liu, G. Dissanayake, J. Valls Miro, and K. Waldron, “Infrastructure robotics: Research challenges and opportunities,” inISARC, 2014

  3. [3]

    Exploration and mapping of catacombs with mobile robots,

    V. A. Ziparo, M. Zaratti, G. Grisetti, T. M. Bonanni, J. Serafin, M. Di Cicco, M. Proesmans, L. Van Gool, O. Vysotska, I. Bogoslavskyi et al., “Exploration and mapping of catacombs with mobile robots,” in SSRR, 2013

  4. [4]

    A 3d mixed reality interface for human-robot teaming,

    J. Chen, B. Sun, M. Pollefeys, and H. Blum, “A 3d mixed reality interface for human-robot teaming,” inICRA, 2024

  5. [5]

    Aerial image-based inter-day registration for precision agriculture,

    C. Gao, F. Daxinger, L. Roth, F. Maffra, P. Beardsley, M. Chli, and L. Teixeira, “Aerial image-based inter-day registration for precision agriculture,” inICRA, 2024

  6. [6]

    Estimating 4d data associations towards spatial-temporal mapping of growing plants for agricultural robots,

    L. Lobefaro, M. V. Malladi, O. Vysotska, T. Guadagnino, and C. Stach- niss, “Estimating 4d data associations towards spatial-temporal mapping of growing plants for agricultural robots,” inIROS, 2023

  7. [7]

    Objectnav revisited: On evaluation of embodied agents navigating to objects,

    D. Batra, A. Gokaslan, A. Kembhavi, O. Maksymets, R. Mottaghi, M. Savva, A. Toshev, and E. Wijmans, “Objectnav revisited: On evaluation of embodied agents navigating to objects,”arXiv preprint arXiv:2006.13171, 2020

  8. [8]

    A frontier-based approach for autonomous exploration,

    B. Yamauchi, “A frontier-based approach for autonomous exploration,” in International Symposium on Computational Intelligence in Robotics and Automation CIRA, 1997

Show all 44 references
  1. [9]

    Receding horizon

    A. Bircher, M. Kamel, K. Alexis, H. Oleynikova, and R. Siegwart, “Receding horizon" next-best-view" planner for 3d exploration,” in ICRA, 2016

  2. [10]

    Vlfm: Vision- language frontier maps for zero-shot semantic navigation,

    N. Yokoyama, S. Ha, D. Batra, J. Wang, and B. Bucher, “Vlfm: Vision- language frontier maps for zero-shot semantic navigation,” in ICRA, 2024

  3. [11]

    An efficient sampling-based method for online informative path planning in unknown environments,

    L. Schmid, M. Pantic, R. Khanna, L. Ott, R. Siegwart, and J. Nieto, “An efficient sampling-based method for online informative path planning in unknown environments,”IEEE Robotics and Automation Letters, 2020

  4. [12]

    Fast frontier-based information-driven autonomous exploration with an mav,

    A. Dai, S. Papatheodorou, N. Funk, D. Tzoumanikas, and S. Leuteneg- ger, “Fast frontier-based information-driven autonomous exploration with an mav,” inICRA, 2020

  5. [13]

    Finding things in the unknown: Semantic object-centric exploration with an mav,

    S. Papatheodorou, N. Funk, D. Tzoumanikas, C. Choi, B. Xu, and S. Leutenegger, “Finding things in the unknown: Semantic object-centric exploration with an mav,” inICRA, 2023

  6. [14]

    Fuel: Fast uav exploration using incremental frontier structure and hierarchical planning,

    B. Zhou, Y. Zhang, X. Chen, and S. Shen, “Fuel: Fast uav exploration using incremental frontier structure and hierarchical planning,” IEEE Robotics and Automation Letters, vol. 6, no. 2, pp. 779–786, 2021

  7. [15]

    Seer:Safeefficientexplorationforaerialrobotsusinglearningtopredict information gain,

    Y. Tao, Y. Wu, B. Li, F. Cladera, A. Zhou, D. Thakur, and V. Kumar, “Seer:Safeefficientexplorationforaerialrobotsusinglearningtopredict information gain,” inICRA, 2023

  8. [16]

    Active Neural Localization,

    D. S. Chaplot, E. Parisotto, and R. Salakhutdinov, “Active Neural Localization,” ICLR, 2018

  9. [17]

    Tare: A hierarchical frame- work for efficiently exploring complex 3d environments

    C. Cao, H. Zhu, H. Choset, and J. Zhang, “Tare: A hierarchical frame- work for efficiently exploring complex 3d environments.” inRobotics: Science and Systems, vol. 5, 2021, p. 2

  10. [18]

    Efficient submap-based autonomous mav exploration using visual- inertial slam configurable for lidars or depth cameras,

    S. Papatheodorou, S. Boche, S. B. Laina, and S. Leutenegger, “Efficient submap-based autonomous mav exploration using visual- inertial slam configurable for lidars or depth cameras,”arXiv preprint arXiv:2409.16972, 2024

  11. [19]

    An improved frontier-based approach for autonomous exploration,

    W. Gao, M. Booker, A. Adiwahono, M. Yuan, J. Wang, and Y. W. Yun, “An improved frontier-based approach for autonomous exploration,” in 2018 15th international conference on control, automation, robotics and vision (ICARCV). IEEE, 2018, pp. 292–297

  12. [20]

    Efficient autonomous exploration planning of large-scale 3-d environments,

    M. Selin, M. Tiger, D. Duberg, F. Heintz, and P. Jensfelt, “Efficient autonomous exploration planning of large-scale 3-d environments,” Robotics and Automation Letters, 2019

  13. [21]

    Re- ceding horizon path planning for 3d exploration and surface inspection,

    A. Bircher, M. Kamel, K. Alexis, H. Oleynikova, and R. Siegwart, “Re- ceding horizon path planning for 3d exploration and surface inspection,” Autonomous Robots, vol. 42, pp. 291–306, 2018

  14. [22]

    Sc-explorer: Incremental 3d scene completion for safe and efficient exploration mapping and planning,

    L. Schmid, M. N. Cheema, V. Reijgwart, R. Siegwart, F. Tombari, and C. Cadena, “Sc-explorer: Incremental 3d scene completion for safe and efficient exploration mapping and planning,” arXiv preprint arXiv:2208.08307, 2022

  15. [23]

    Active neural mapping,

    Z. Yan, H. Yang, and H. Zha, “Active neural mapping,” inICCV, 2023

  16. [24]

    Uncertainty guided policy for active robotic 3d reconstruction using neural radiance fields,

    S. Lee, C. Le, W. Jiahao, A. Liniger, S. Kumar, and F. Yu, “Uncertainty guided policy for active robotic 3d reconstruction using neural radiance fields,” IEEE Robotics and Automation Letters, 2022

  17. [25]

    Fisherrf: Active view selection and uncertainty quantification for radiance fields using fisher information,

    W. Jiang, B. Lei, and K. Daniilidis, “Fisherrf: Active view selection and uncertainty quantification for radiance fields using fisher information,” arXiv, 2023

  18. [26]

    Ag-slam: Active gaussian splatting slam,

    W. Jiang, B. Lei, K. Ashton, and K. Daniilidis, “Ag-slam: Active gaussian splatting slam,”arXiv preprint arXiv:2410.17422, 2024

  19. [27]

    Rt-guide: Real-time gaussian splatting for information-driven explo- ration,

    Y. Tao, D. Ong, V. Murali, I. Spasojevic, P. Chaudhari, and V. Kumar, “Rt-guide: Real-time gaussian splatting for information-driven explo- ration,” arXiv preprint arXiv:2409.18122, 2024

  20. [28]

    Active bayesian multi-class map- ping from range and semantic segmentation observations,

    A. Asgharivaskasi and N. Atanasov, “Active bayesian multi-class map- ping from range and semantic segmentation observations,” in ICRA, 2021

  21. [29]

    3d active metric-semantic slam,

    Y. Tao, X. Liu, I. Spasojevic, S. Agarwal, and V. Kumar, “3d active metric-semantic slam,”IEEE Robotics and Automation Letters, 2024

  22. [30]

    Seal: Self-supervised embodied active learning using exploration and 3d consistency,

    D. S. Chaplot, M. Dalal, S. Gupta, J. Malik, and R. R. Salakhutdinov, “Seal: Self-supervised embodied active learning using exploration and 3d consistency,”NeurIPS, 2021

  23. [31]

    How to not train your dragon: Training-free embodied object goal navigation with semantic frontiers,

    J. Chen, G. Li, S. Kumar, B. Ghanem, and F. Yu, “How to not train your dragon: Training-free embodied object goal navigation with semantic frontiers,” arXiv preprint arXiv:2305.16925, 2023

  24. [32]

    Ippon: Common sense guided informative path planning for object goal navi- gation,

    K. Qu, J. Tan, T. Zhang, F. Xia, C. Cadena, and M. Hutter, “Ippon: Common sense guided informative path planning for object goal navi- gation,” arXiv preprint arXiv:2410.19697, 2024

  25. [33]

    Metric3d v2: A versatile monocular geometric foundation model for zero-shot metric depth and surface normal estimation,

    M. Hu, W. Yin, C. Zhang, Z. Cai, X. Long, H. Chen, K. Wang, G. Yu, C. Shen, and S. Shen, “Metric3d v2: A versatile monocular geometric foundation model for zero-shot metric depth and surface normal estimation,”arXiv preprint arXiv:2404.15506, 2024

  26. [34]

    Deeplsd: Line segment detection and refinement with deep image gradients,

    R. Pautrat, D. Barath, V. Larsson, M. R. Oswald, and M. Pollefeys, “Deeplsd: Line segment detection and refinement with deep image gradients,” inCVPR, 2023

  27. [35]

    Holistically-attracted wireframe parsing,

    N. Xue, T. Wu, S. Bai, F. Wang, G.-S. Xia, L. Zhang, and P. H. Torr, “Holistically-attracted wireframe parsing,” inCVPR, 2020

  28. [36]

    Adabins: Depth estimation using adaptive bins,

    S. F. Bhat, I. Alhashim, and P. Wonka, “Adabins: Depth estimation using adaptive bins,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 4009–4018

  29. [37]

    Habitat-matterport 3d dataset (hm3d): 1000 large-scale 3d environments for embodied ai,

    S. K. Ramakrishnan, A. Gokaslan, E. Wijmans, O. Maksymets, A. Clegg, J. Turner, E. Undersander, W. Galuba, A. Westbury, A. X. Chang et al., “Habitat-matterport 3d dataset (hm3d): 1000 large-scale 3d environments for embodied ai,”arXiv:2109.08238, 2021

  30. [38]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inCVPR, 2016

  31. [39]

    Imagenet large scale visual recognition challenge,

    O.Russakovsky,J.Deng,H.Su,J.Krause,S.Satheesh,S.Ma,Z.Huang, A. Karpathy, A. Khosla, M. Bernsteinet al., “Imagenet large scale visual recognition challenge,”IJCV, 2015

  32. [40]

    A unified approach for autonomous volumetric exploration of large scale environments under severe odometry drift,

    L. Schmid, V. Reijgwart, L. Ott, J. Nieto, R. Siegwart, and C. Cadena, “A unified approach for autonomous volumetric exploration of large scale environments under severe odometry drift,”IEEE Robotics and Automation Letters, vol. 6, no. 3, pp. 4504–4511, 2021

  33. [41]

    Density-based clustering based on hierarchical density estimates,

    R. J. Campello, D. Moulavi, and J. Sander, “Density-based clustering based on hierarchical density estimates,” inPacific-Asia conference on knowledge discovery and data mining. Springer, 2013, pp. 160–172

  34. [42]

    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

  35. [43]

    OctoMap: An efficient probabilistic 3D mapping framework based on octrees,

    A. Hornung, K. M. Wurm, M. Bennewitz, C. Stachniss, and W. Burgard, “OctoMap: An efficient probabilistic 3D mapping framework based on octrees,” Autonomous Robots, 2013

  36. [44]

    The Open Motion Planning Library,

    I. A. Şucan, M. Moll, and L. E. Kavraki, “The Open Motion Planning Library,” IEEE Robotics & Automation Magazine, 2012

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.