Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

ActiveGAMER: Active GAussian Mapping through Efficient Rendering

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read ActiveGAMER achieves best-reported active mapping and rendering by planning next views with a rendering-based information gain computed from a 3D Gaussian Splatting map.

desk verdict ActiveGAMER is a solid 3DGS-based active mapping system with honest ablations and a real-time rendering advantage, but its SOTA claim rests on two different checkpoints and a limited baseline set; a single-checkpoint re-evaluation would firm it up. read the letter →

arxiv 2501.06897 v3 pith:QGJWDWXB submitted 2025-01-12 cs.CV cs.RO

Liyan Chen , Huangying Zhan , Kevin Chen , Xiangyu Xu , Qingan Yan , Changjiang Cai , Yi Xu This is my paper · ORCID
classification cs.CVcs.RO
keywords activemapping3DGaussianSplattingnext-best-viewplanninginformationgainreconstructionsceneexplorationnovelviewsynthesis
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

ActiveGAMER claims that 3D Gaussian Splatting, not neural radiance fields, is the right backbone for active mapping: because the Gaussian map renders posed RGB-D frames in real time, the system can score a large pool of candidate viewpoints by how many previously unseen pixels each would reveal. The paper builds a full exploration loop around that score, adding coarse-to-fine candidate sampling, a global-local keyframe schedule, and a post-refinement pass. On the Replica and Matterport3D benchmarks the system reports the best geometric accuracy and completeness among compared active mappers, and the best novel-view rendering quality. If the claim holds, active reconstruction stops being a slow offline process and becomes a real-time loop that can run on a moving camera.

What carries the argument

The load-bearing object is the rendering-based information gain computed from the Gaussian Map's silhouette mask $S(p)=\sum_i f_i(p)\prod_{j<i}(1-f_j(p))$, which says whether a pixel has received any Gaussian contribution. The exploration score $I=(1-\sigma(l_i))\,\sigma(\log N_{S_i})$ combines the count $N_{S_i}$ of zero-silhouette pixels at a candidate pose with a softmax travel-distance penalty, so the planner prefers nearby views that reveal the most currently invisible map area. This turns candidate evaluation into one cheap render per hypothesis, which is what makes the 6DoF active search practical on a V100-class GPU.

What would settle it

Take a scene with a large thin double-sided panel and run ActiveGAMER with a fixed step budget; under Eq. (8), a pose aimed at the unseen back side scores near-zero information gain because it renders no new missing pixels, so if the back side nonetheless gets reconstructed, the planning signal is not the cause. A direct check is to log the scores: the back-side view should be the last or never selected even though it adds genuinely new surface.

Watch

Extended reading notes

Core claim

The central discovery is that a rendered silhouette mask from the current Gaussian Map is a sufficient exploration signal. For each candidate pose, ActiveGAMER renders the map, counts zero pixels in the silhouette (missing coverage), and scores the view by $I=(1-\sigma(l_i))\cdot\sigma(\log(N_{S_i}))$, where $l_i$ is travel distance; the candidate with the highest score becomes the next goal. Because 3DGS renders fast, many candidates can be evaluated repeatedly during a run, something NeRF-based mappers cannot afford. With this planner plus the keyframe and refinement components, the paper reports 1.66 cm accuracy, 2.30 cm completion, and 95.32% completion ratio at 5 cm on MP3D, and 32.02 PSNR, 0.97 SSIM, 0.11 LPIPS on Replica novel views, surpassing the compared active mappers on both geometry and rendering.

Load-bearing premise

The assumption that carries the system is that the number of unseen pixels in a rendered silhouette is a faithful proxy for how much a candidate view will improve reconstruction; the paper itself reports cases, such as double-sided objects and wall-hugging viewpoints, where that proxy misses real gains.

Editorial extensions

If this is right

  • Active mapping on a 3D Gaussian Splatting backbone runs at interactive speeds, so a robot can re-plan next-best views while mapping rather than in a separate offline stage.
  • Counting missing pixels in a rendered silhouette is enough to guide exploration in indoor scenes, removing the need for learned uncertainty estimators or volumetric occupancy grids as the planning signal.
  • Unrestricted 6DoF candidate sampling, combined with coarse-to-fine search, reconstructs scenes more completely than 2D-plane or hemisphere-constrained NeRF planners.
  • The explicit Gaussian point cloud avoids the extrapolation artifacts of neural maps, so geometry evaluation can be read directly from the map rather than from a separately reconstructed surface.
  • Optimizing the Gaussian map for rendering can prune geometry; evaluating geometry on the exploration model and rendering on the refined model gives a clean way to report both.

Reading between the lines

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

  • Beyond the paper: the same zero-silhouette-pixel signal could be computed from any differentiable renderer, so the planning module may transfer to mesh or SDF-based maps without retraining.
  • Beyond the paper: adding semantic or surface cues to the information gain, as the paper itself suggests, would likely fix the double-sided-object failure; a cheap test is to penalize views whose rendered normal map matches already observed normals.
  • Beyond the paper: because the method assumes known poses and perfect action execution, the next practical step is to couple planning with an active SLAM objective, trading off map coverage against pose uncertainty.
  • Beyond the paper: the missing-pixel count could serve as a reward signal for a learned exploration policy, converting the hand-designed softmax scoring into a differentiable or reinforcement-learned planner.
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

4 major / 5 minor

Summary. This manuscript presents ActiveGAMER, an active 3D Gaussian Splatting (3DGS) mapping system that combines a rendering-based information gain for next-best-view selection with coarse-to-fine exploration, global-local keyframe selection, and post-refinement. The authors report state-of-the-art geometric reconstruction on MP3D (Accuracy 1.66 cm, Completion 2.30 cm, Completion Ratio 95.32%) and novel-view rendering on Replica (PSNR 32.02, SSIM 0.97, LPIPS 0.11) compared to NeRF-based active mapping baselines. The system runs in real time and uses a Gaussian Map backbone (SplaTAM).

Significance. If the results are reproducible under a consistent evaluation protocol, the paper demonstrates a practical advantage of 3DGS for active mapping, particularly in photometric quality and exploration efficiency. The rendering-based information gain is a simple and computationally cheap planning signal that leverages the differentiable renderer. However, the evaluation protocol currently undermines the headline SOTA claim, and the lack of numerical comparison with concurrent 3DGS active mappers leaves the contribution's relative position unclear.

major comments (4)
  1. [Sec. 4.2, Tables 1 and 2] The geometric metrics are computed on the Exploration Model, while rendering metrics are computed on the Refinement Model. Since post-refinement is described in Sec. 3.4 as part of the pipeline and can degrade geometry, the reported numbers do not describe a single system output. The authors should report both metrics from the same model (ideally the final output after post-refinement), or explicitly justify why the two-model evaluation is the fair representation of the active mapping system's performance.
  2. [Sec. 4.2, Table 2] The evaluation masks for the two baselines are inconsistent: SplaTAM excludes uncovered regions while NARUTO uses all pixels, and the mask used for "Ours" is not stated. This makes the PSNR/SSIM/LPIPS comparison ambiguous and potentially unfair. Specify the mask for all methods and use a common protocol (e.g., all pixels within the explored free space, or all pixels for all methods with the same definition of coverage).
  3. [Sec. 2 and Sec. 4.2] The paper cites concurrent 3DGS-based active mapping systems (ActiveSplat [33], AG-SLAM [24]) but does not provide any quantitative comparison against them. Given the paper's claim to "significantly surpass" existing approaches, numerical or at least qualitative comparison with these directly related methods is necessary to substantiate the SOTA claim.
  4. [Eq. (8) and Sec. 9.4] The information gain is defined as the number of zero pixels in the silhouette mask rendered from the current Gaussian Map. The authors themselves identify structural cases where this proxy fails (double-sided objects, near-wall candidates). While these are acknowledged as future work, the paper should provide evidence about how frequently such cases occur in the benchmark scenes and how much they affect the final reconstruction. A concrete test would be comparing the planning decisions with ground-truth information gain (e.g., computed from the GT mesh) on a subset of steps.
minor comments (5)
  1. [Sec. 3.2, Eq. (8)] The notation N_{S_i} = \sum_p I(S_i(p) = 0) is ambiguous because the indicator function is not explicitly defined; please use a distinct symbol such as [S_i(p) = 0] or a function name to denote the indicator.
  2. [Sec. 4.2] The phrase "among the first to address active mapping using 3D Gaussian Splatting" is contradicted by the concurrent works cited in the same paragraph; please rephrase as "one of the first" and cite those works more explicitly.
  3. [Sec. 3.3] The Type-2 global keyframe selection criterion "rendering quality falls below a threshold" does not specify the threshold or the quality metric; please provide the concrete value used in experiments.
  4. [Sec. 4.2] The description of the SplaTAM baseline says "disable its tracking thread and simulate handheld scanning" but the scanning trajectory is not defined; please describe the trajectory to allow reproducibility.
  5. [Table 3] The "Full" row shows higher PSNR but worse Completion than "w/o Refine"; the caption should clarify whether the reported geometry is from the Exploration Model or the final model for each ablation.

Circularity Check

0 steps flagged · score 2.0 of 10

No derivation-chain circularity: the information-gain planner is a closed-loop heuristic and the SOTA numbers are external-benchmark comparisons; the two-checkpoint evaluation and omitted concurrent baselines are benchmarking concerns, not circularity.

full rationale

ActiveGAMER's claimed chain is empirical rather than derived: it builds a Gaussian map from posed RGB-D frames, selects viewpoints by a rendering-based information gain (Eq. 8), and evaluates against external ground-truth geometry and novel views. The information gain counts missing silhouette pixels from the current map, so it is a feedback control signal, not a quantity forced to equal the reported Acc/Comp/PSNR numbers; the reported metrics come from independent GT-mesh and GT-image comparisons. Hyperparameters such as lambda = 50 x MDE, the 0.5% removal threshold, and the 10% keyframe-novelty threshold are disclosed hand-set constants rather than fitted values renamed as predictions. The paper cites NARUTO (overlapping authorship) for its RRT planner and as a baseline, and ActiveRMAP in related work, but these citations are not load-bearing for the central claim: the planner module is a minor component, and the baseline numbers are externally evaluated. The authors also document genuine limitations of the planning proxy (double-sided objects, near-wall regions, Sec. 5 and Sec. 9.4), which are failure modes of a heuristic, not circularity. The main validity concerns are the separate checkpoints used for geometry (Exploration Model) versus rendering (Refinement Model) in Sec. 4.2 and the absence of numerical comparison to concurrent 3DGS active mappers (ActiveSplat, AG-SLAM), but these affect how the SOTA claim should be read rather than showing that any result is equivalent to its inputs by construction.

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

The system's central behavior depends on several hand-set thresholds (lambda = 50 times MDE, 0.5% candidate removal, 10% keyframe novelty, coarse/fine sampling densities, iteration counts) and on strong domain assumptions: known pose, simulator-perfect observations, and the unvalidated proxy that silhouette zero-pixels measure information gain. No fitted constants are disguised as predictions; the thresholds are disclosed but not systematically tuned or justified by sensitivity analysis. No new physical entities are introduced.

free parameters (6)
  • Densification depth-error threshold factor lambda_MDE = 50
    In Eq. 7, new Gaussians are added where L1(D(p)) > lambda * MDE; chosen as 50 times the median depth error, hand-set.
  • Candidate removal threshold = 0.5% of pixels
    Candidates with N_Si below 0.5% of total pixels are dropped from the exploration pool; hand-set.
  • Global keyframe novelty threshold = 10% new pixels
    Type-1 global keyframes require revealing over 10% new silhouette pixels; hand-set.
  • Coarse/fine candidate sampling densities = coarse v1=1 m, v2=5; fine v1=0.5 m, v2=15
    Spatial step and number of viewing directions per voxel in the exploration candidate pool; chosen by the authors.
  • Optimization iterations and resolution = 15 iterations at quarter resolution; post-refinement 60 at full resolution
    Exploration uses 15 iterations at quarter resolution, refinement 60 at full resolution; hand-set to balance speed and quality.
  • Type-2 global keyframe quality threshold
    A keyframe qualifies as global if its rendering quality 'falls below a threshold', but the threshold value is not specified in the text.
assumptions (5)
  • standard math 3D Gaussian splatting rendering equations (alpha compositing, Eq. 2-5) are correct and inherited from prior work
    The paper's losses and planning rely on this rendering model without re-deriving it; it is standard 3DGS background.
  • domain assumption Known camera poses and perfect action execution
    The system assumes localization is known and actions are executed exactly; stated in Sec. 1 and Sec. 5 as limiting future work.
  • domain assumption The simulator provides realistic posed RGB-D data
    All experiments run in Habitat Simulator on Replica and MP3D; results on real sensors would require additional robustness.
  • ad hoc to paper Missing-pixel count in the rendered silhouette mask is a monotone proxy for reconstruction information gain
    Eq. 8 defines information gain I in terms of N_Si, the count of zero pixels; the paper provides no derivation or independent validation, and Sec. 9.4 identifies counterexamples.
  • domain assumption SplaTAM-style Gaussian mapping is a sufficient backbone for active mapping
    The mapping module inherits SplaTAM [25] wholesale; its known limitations, such as Gaussian pruning degrading geometry, directly shape the results.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ActiveGAMER: Active GAussian Mapping through Efficient Rendering." pith.science (2026). https://pith.science/paper/QGJWDWXB

@misc{pith2026250106897,
  author       = {Pith},
  title        = {Pith review of: ActiveGAMER: Active GAussian Mapping through Efficient Rendering},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QGJWDWXB}},
  note         = {Machine review of arXiv:2501.06897}
}
read the original abstract

We introduce ActiveGAMER, an active mapping system that utilizes 3D Gaussian Splatting (3DGS) to achieve high-quality, real-time scene mapping and exploration. Unlike traditional NeRF-based methods, which are computationally demanding and restrict active mapping performance, our approach leverages the efficient rendering capabilities of 3DGS, allowing effective and efficient exploration in complex environments. The core of our system is a rendering-based information gain module that dynamically identifies the most informative viewpoints for next-best-view planning, enhancing both geometric and photometric reconstruction accuracy. ActiveGAMER also integrates a carefully balanced framework, combining coarse-to-fine exploration, post-refinement, and a global-local keyframe selection strategy to maximize reconstruction completeness and fidelity. Our system autonomously explores and reconstructs environments with state-of-the-art geometric and photometric accuracy and completeness, significantly surpassing existing approaches in both aspects. Extensive evaluations on benchmark datasets such as Replica and MP3D highlight ActiveGAMER's effectiveness in active mapping tasks.

Figures

Figures reproduced from arXiv: 2501.06897 by the authors.

Figure 1
Figure 1. ActiveGAMER is built on a Gaussian Map backbone and autonomously performs coarse-to-fine exploration to optimize geometric accuracy and photometric fidelity. known as Active Reconstruction, where localization is as￾sumed to be known, allowing the system to prioritize high￾quality and complete scene reconstruction. We explore this task by introducing a novel approach based on 3D Gaussian Splatting (3DGS), which provi… view at source ↗
Figure 2
Figure 2. ActiveGAMER Framework. At each keyframe step, HabitatSim [52] generates posed RGB-D images, which are stored in a keyframe database, with certain frames designated as Global Keyframes. These observations are used to update a Gaussian Map comprising a collection of 3D Gaussians. Map optimization is achieved by minimizing color and depth rendering losses, based on the rendered RGB-D images and silhouette masks. Using … view at source ↗
Figure 3
Figure 3. 3D Reconstruction Results on MP3D. Shown are two scenes (Left: pLe4; Right: HxpK) with results distinguished by border colors: [Ground Truth, NARUTO [18], Ours]. For NARUTO, black regions indicate neural mapping extrapolation, sometimes with inaccuracies. For our method, colored point clouds are extracted from the Gaussian Map for visualization; note that noisy points here do not reflect actual rendering quality. Un… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Rendering Results on Replica. Two scenes (office0, office4) are shown in the first and second rows, respectively. The results represent [Ground Truth, NARUTO, SplaTAM w/o tracking, Ours]. Our renders demonstrate higher fidelity across most regions. Fine-Exploration Sta…
Figure 5
Figure 5. Figure 5 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Novel View Rendering Results on Replica. This figure provides a side-by-side comparison of rendering results on the Replica dataset. Upper rows show the ground truth RGB-D images, while bottom rows present the rendered RGB-D images. comprehensive visualization, we pres…
Figure 7
Figure 7. Figure 7: MP3D Reconstruction and Rendering Results. This figure provides a side-by-side comparison of reconstruction results on the MP3D dataset. Odd-numbered rows show the ground truth meshes and RGB-D images, while even-numbered rows present the 3D point cloud extracted from …
Figure 8
Figure 8. Figure 8: The exploration trajectories for selected scenes in Replica are visualized, showing only keyframe cameras. Camera colors indicate [PITH_FULL_IMAGE:figures/full_fig_p018_8.png]
Figure 9
Figure 9. Figure 9: Reconstruction Failure Cases. Our method encounters two common failure scenarios: (A) Insufficient observations due to ignored candidates near surface regions. (B) Unreconstructed back sides of double-sided objects, as these regions are not captured by the rendering-ba…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. JointSplat: Probabilistic Joint Flow-Depth Optimization for Sparse-View Gaussian Splatting

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A feed-forward 3D Gaussian splatting method fuses depth and optical flow via a learned reliability mask, improving novel-view PSNR on RealEstate10K by 0.19 dB over its backbone.

  2. GauSS-MI: Gaussian Splatting Shannon Mutual Information for Active 3D Reconstruction

    cs.GR 2025-04 reject novelty 6.0 of 10

    GauSS-MI estimates the expected visual information gain of a viewpoint from per-Gaussian reliability probabilities, and an active reconstruction system using it achieves high visual fidelity with fewer frames.

Reference graph

Works this paper leans on

80 extracted references · 71 canonical work pages · cited by 2 Pith papers

  1. [33]

    Activesplat: High-fidelity scene reconstruction through active gaussian splatting

    Yuetao Li, Zijia Kuang, Ting Li, Guyue Zhou, Shao- hui Zhang, and Zike Yan. Activesplat: High-fidelity scene reconstruction through active gaussian splatting. arXiv preprint arXiv:2410.21955, 2024. 3, 6

    arXiv 2024

  2. [24]

    Ag-slam: Active gaussian splatting slam

    Wen Jiang, Boshu Lei, Katrina Ashton, and Kostas Daniilidis. Ag-slam: Active gaussian splatting slam. arXiv preprint arXiv:2410.17422, 2024. 3, 6

    arXiv 2024

  3. [1]

    Vision-only robot naviga- tion in a neural radiance world.IEEE Robotics and Automation Letters, 7(2):4606–4613, 2022

    Michal Adamkiewicz, Timothy Chen, Adam Caccav- ale, Rachel Gardner, Preston Culbertson, Jeannette Bohg, and Mac Schwager. Vision-only robot naviga- tion in a neural radiance world.IEEE Robotics and Automation Letters, 7(2):4606–4613, 2022. 2

    work page 2022

  4. [2]

    Neural rgb-d surface reconstruction

    Dejan Azinovi ´c, Ricardo Martin-Brualla, Dan B Gold- man, Matthias Nießner, and Justus Thies. Neural rgb-d surface reconstruction. InProceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 6290–6301, 2022. 1, 2

    work page 2022

  5. [3]

    Unsupervised scale-consistent depth and ego- motion learning from monocular video.Advances in neural information processing systems, 32, 2019

    Jiawang Bian, Zhichao Li, Naiyan Wang, Huangy- ing Zhan, Chunhua Shen, Ming-Ming Cheng, and Ian Reid. Unsupervised scale-consistent depth and ego- motion learning from monocular video.Advances in neural information processing systems, 32, 2019. 2

    work page 2019

  6. [4]

    Information based adaptive robotic explo- ration

    Frederic Bourgault, Alexei A Makarenko, Stefan B Williams, Ben Grocholsky, and Hugh F Durrant- Whyte. Information based adaptive robotic explo- ration. InIEEE/RSJ international conference on in- telligent robots and systems, pages 540–545. IEEE,

    work page

  7. [5]

    Past, present, and future of simulta- neous localization and mapping: Toward the robust- perception age.IEEE Transactions on robotics, 32(6): 1309–1332, 2016

    Cesar Cadena, Luca Carlone, Henry Carrillo, Yasir Latif, Davide Scaramuzza, Jos ´e Neira, Ian Reid, and John J Leonard. Past, present, and future of simulta- neous localization and mapping: Toward the robust- perception age.IEEE Transactions on robotics, 32(6): 1309–1332, 2016. 2

    work page 2016

  8. [6]

    Riav- mvs: Recurrent-indexing an asymmetric volume for multi-view stereo

    Changjiang Cai, Pan Ji, Qingan Yan, and Yi Xu. Riav- mvs: Recurrent-indexing an asymmetric volume for multi-view stereo. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, pages 919–928, 2023. 2

    work page 2023

Show all 80 references
  1. [7]

    Matter- port3d: Learning from rgb-d data in indoor environ- ments.arXiv preprint arXiv:1709.06158, 2017

    Angel Chang, Angela Dai, Thomas Funkhouser, Ma- ciej Halber, Matthias Niessner, Manolis Savva, Shu- ran Song, Andy Zeng, and Yinda Zhang. Matter- port3d: Learning from rgb-d data in indoor environ- ments.arXiv preprint arXiv:1709.06158, 2017. 6, 3

    2017 arXiv

  2. [8]

    Learning to explore using active neural slam.arXiv preprint arXiv:2004.05155, 2020

    Devendra Singh Chaplot, Dhiraj Gandhi, Saurabh Gupta, Abhinav Gupta, and Ruslan Salakhutdinov. Learning to explore using active neural slam.arXiv preprint arXiv:2004.05155, 2020. 6

    2004 arXiv

  3. [9]

    Learning the distribution of errors in stereo matching for joint disparity and uncertainty estimation

    Liyan Chen, Weihan Wang, and Philippos Mordohai. Learning the distribution of errors in stereo matching for joint disparity and uncertainty estimation. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17235–17244,

  4. [10]

    Gaussian activated neural ra- diance fields for high fidelity reconstruction and pose estimation

    Shin-Fang Chng, Sameera Ramasinghe, Jamie Sher- rah, and Simon Lucey. Gaussian activated neural ra- diance fields for high fidelity reconstruction and pose estimation. InEuropean Conference on Computer Vi- sion, pages 264–280. Springer, 2022. 2

    2022

  5. [11]

    The determination of next best views

    Cl Connolly. The determination of next best views. In Proceedings. 1985 IEEE international conference on robotics and automation, pages 432–435. IEEE, 1985. 2

    1985

  6. [12]

    Andrew J Davison and David W. Murray. Simultane- ous localization and map-building using active vision. IEEE transactions on pattern analysis and machine intelligence, 24(7):865–880, 2002. 2

    2002

  7. [13]

    Monoslam: Real-time single cam- era slam.IEEE transactions on pattern analysis and machine intelligence, 29(6):1052–1067, 2007

    Andrew J Davison, Ian D Reid, Nicholas D Molton, and Olivier Stasse. Monoslam: Real-time single cam- era slam.IEEE transactions on pattern analysis and machine intelligence, 29(6):1052–1067, 2007. 2

    2007

  8. [14]

    A comparison of volumetric in- formation gain metrics for active 3d object reconstruc- tion.Autonomous Robots, 42(2):197–208, 2018

    Jeffrey Delmerico, Stefan Isler, Reza Sabzevari, and Davide Scaramuzza. A comparison of volumetric in- formation gain metrics for active 3d object reconstruc- tion.Autonomous Robots, 42(2):197–208, 2018. 2

    2018

  9. [15]

    Simultaneous localization and mapping: part i.IEEE robotics & automation magazine, 13(2):99–110, 2006

    Hugh Durrant-Whyte and Tim Bailey. Simultaneous localization and mapping: part i.IEEE robotics & automation magazine, 13(2):99–110, 2006. 2

    2006

  10. [16]

    Depth map prediction from a single image using a multi-scale deep network.Advances in neural infor- mation processing systems, 27, 2014

    David Eigen, Christian Puhrsch, and Rob Fergus. Depth map prediction from a single image using a multi-scale deep network.Advances in neural infor- mation processing systems, 27, 2014. 2

    2014

  11. [17]

    Adaptive mobile robot navigation and mapping.The International Journal of Robotics Re- search, 18(7):650–668, 1999

    Hans Jacob S Feder, John J Leonard, and Christo- pher M Smith. Adaptive mobile robot navigation and mapping.The International Journal of Robotics Re- search, 18(7):650–668, 1999. 2

    1999

  12. [18]

    Naruto: Neural active reconstruc- tion from uncertain target observations

    Ziyue Feng, Huangying Zhan, Zheng Chen, Qingan Yan, Xiangyu Xu, Changjiang Cai, Bing Li, Qilun Zhu, and Yi Xu. Naruto: Neural active reconstruc- tion from uncertain target observations. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, page...

    2024

  13. [19]

    Uncertainty-driven planner for exploration and navigation

    Georgios Georgakis, Bernadette Bucher, Anton Ara- pin, Karl Schmeckpeper, Nikolai Matni, and Kostas Daniilidis. Uncertainty-driven planner for exploration and navigation. In2022 International Conference on Robotics and Automation (ICRA), pages 11295– 11302. IEEE, 2022. 6

    2022

  14. [20]

    A for- mal basis for the heuristic determination of minimum cost paths.IEEE Transactions on Systems Science and Cybernetics, 4(2):100–107, 1968

    Peter Hart, Nils Nilsson, and Bertram Raphael. A for- mal basis for the heuristic determination of minimum cost paths.IEEE Transactions on Systems Science and Cybernetics, 4(2):100–107, 1968. 2

    1968

  15. [21]

    Accurate and efficient stereo pro- cessing by semi-global matching and mutual informa- tion

    Heiko Hirschmuller. Accurate and efficient stereo pro- cessing by semi-global matching and mutual informa- tion. In2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), pages 807–814. IEEE, 2005. 2

    2005

  16. [22]

    2d gaussian splatting for geometrically accurate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2d gaussian splatting for geometrically accurate radiance fields. InACM SIG- GRAPH 2024 Conference Papers, pages 1–11, 2024. 2

    2024

  17. [23]

    An information gain formulation for active volumetric 3d reconstruction

    Stefan Isler, Reza Sabzevari, Jeffrey Delmerico, and Davide Scaramuzza. An information gain formulation for active volumetric 3d reconstruction. In2016 IEEE International Conference on Robotics and Automation (ICRA), pages 3477–3484. IEEE, 2016. 2

    2016

  18. [25]

    Splatam: Splat track & map 3d gaussians for dense rgb-d slam

    Nikhil Keetha, Jay Karhade, Krishna Murthy Jataval- labhula, Gengshan Yang, Sebastian Scherer, Deva Ra- manan, and Jonathon Luiten. Splatam: Splat track & map 3d gaussians for dense rgb-d slam. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recogniti...

    2024

  19. [26]

    3d gaussian splatting for real-time radiance field rendering.ACM Trans

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering.ACM Trans. Graph., 42(4):139–1, 2023. 1, 2, 3

    2023

  20. [27]

    Efficient next-best-scan planning for autonomous 3d surface reconstruction of unknown ob- jects.Journal of Real-Time Image Processing, 10(4): 611–631, 2015

    Simon Kriegel, Christian Rink, Tim Bodenm ¨uller, and Michael Suppa. Efficient next-best-scan planning for autonomous 3d surface reconstruction of unknown ob- jects.Journal of Real-Time Image Processing, 10(4): 611–631, 2015. 2

    2015

  21. [28]

    Active neural mapping at scale.arXiv preprint arXiv:2409.20276, 2024

    Zijia Kuang, Zike Yan, Hao Zhao, Guyue Zhou, and Hongbin Zha. Active neural mapping at scale.arXiv preprint arXiv:2409.20276, 2024. 2, 3

    2024 arXiv

  22. [29]

    Rrt-connect: An efficient approach to single-query path plan- ning

    James J Kuffner and Steven M LaValle. Rrt-connect: An efficient approach to single-query path plan- ning. InProceedings 2000 ICRA. Millennium Con- ference. IEEE International Conference on Robotics and Automation. Symposia Proceedings (Cat. No. 00CH37065), pages 995–1001. IEEE...

    2000

  23. [30]

    Rapidly-exploring random trees: Progress and prospects.Algorithmic and computational robotics: new directions, 5:293–308, 2001

    Steven M LaValle, James J Kuffner, BR Donald, et al. Rapidly-exploring random trees: Progress and prospects.Algorithmic and computational robotics: new directions, 5:293–308, 2001. 2, 5

    2001

  24. [31]

    Uncertainty guided policy for active robotic 3d reconstruction us- ing neural radiance fields.IEEE Robotics and Automa- tion Letters, 2022

    Soomin Lee, Le Chen, Jiahao Wang, Alexander Lin- iger, Suryansh Kumar, and Fisher Yu. Uncertainty guided policy for active robotic 3d reconstruction us- ing neural radiance fields.IEEE Robotics and Automa- tion Letters, 2022. 2

    2022

  25. [32]

    Bnv-fusion: Dense 3d reconstruction using bi-level neural volume fusion

    Kejie Li, Yansong Tang, Victor Adrian Prisacariu, and Philip HS Torr. Bnv-fusion: Dense 3d reconstruction using bi-level neural volume fusion. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 1, 2

    2022

  26. [34]

    Barf: Bundle-adjusting neural radi- ance fields

    Chen-Hsuan Lin, Wei-Chiu Ma, Antonio Torralba, and Simon Lucey. Barf: Bundle-adjusting neural radi- ance fields. InProceedings of the IEEE/CVF Interna- tional Conference on Computer Vision, pages 5741– 5751, 2021. 2

    2021

  27. [35]

    Planemvs: 3d plane reconstruction from multi-view stereo

    Jiachen Liu, Pan Ji, Nitin Bansal, Changjiang Cai, Qingan Yan, Xiaolei Huang, and Yi Xu. Planemvs: 3d plane reconstruction from multi-view stereo. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, pages 8665– 8675, 2022. 2

    2022

  28. [36]

    Active mapping and robot exploration: A survey.Sen- sors, 21(7):2445, 2021

    Iker Lluvia, Elena Lazkano, and Ander Ansuategi. Active mapping and robot exploration: A survey.Sen- sors, 21(7):2445, 2021. 2

    2021

  29. [37]

    An exper- iment in integrated exploration

    Alexei A Makarenko, Stefan B Williams, Frederic Bourgault, and Hugh F Durrant-Whyte. An exper- iment in integrated exploration. InIEEE/RSJ inter- national conference on intelligent robots and systems, pages 534–539. IEEE, 2002. 2

    2002

  30. [38]

    Gaussian splatting slam

    Hidenobu Matsuki, Riku Murai, Paul HJ Kelly, and Andrew J Davison. Gaussian splatting slam. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18039–18048,

  31. [39]

    Occlusions as a guide for planning the next view.IEEE transactions on pattern analysis and machine intelligence, 15(5): 417–433, 1993

    Jasna Maver and Ruzena Bajcsy. Occlusions as a guide for planning the next view.IEEE transactions on pattern analysis and machine intelligence, 15(5): 417–433, 1993. 2

    1993

  32. [40]

    Nerf: Representing scenes as neural radiance fields for view synthesis.Communications of the ACM, 65(1): 99–106, 2021

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view synthesis.Communications of the ACM, 65(1): 99–106, 2021. 1, 2

    2021

  33. [41]

    Instant neural graphics primitives with a multiresolution hash encoding.ACM Trans

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexander Keller. Instant neural graphics primitives with a multiresolution hash encoding.ACM Trans. Graph., 41(4):102:1–102:15, 2022. 1

    2022

  34. [42]

    Au- tonomous feature-based exploration

    Paul Newman, Michael Bosse, and John Leonard. Au- tonomous feature-based exploration. In2003 IEEE International Conference on Robotics and Automa- tion (Cat. No. 03CH37422), pages 1234–1240. IEEE,

  35. [43]

    Giraffe: Rep- resenting scenes as compositional generative neural feature fields

    Michael Niemeyer and Andreas Geiger. Giraffe: Rep- resenting scenes as compositional generative neural feature fields. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 11453–11464, 2021. 2

    2021

  36. [44]

    Activenerf: Learning where to see with uncertainty estimation

    Xuran Pan, Zihang Lai, Shiji Song, and Gao Huang. Activenerf: Learning where to see with uncertainty estimation. InEuropean Conference on Computer Vi- sion, pages 230–246. Springer, 2022. 2, 6

    2022

  37. [45]

    Deepsdf: Learning continuous signed distance functions for shape representation

    Jeong Joon Park, Peter Florence, Julian Straub, Richard Newcombe, and Steven Lovegrove. Deepsdf: Learning continuous signed distance functions for shape representation. InProceedings of the IEEE/CVF conference on computer vision and pattern recogni- tion, pages 165–174, 2019. 1, 2

    2019

  38. [46]

    Next-best view policy for 3d reconstruction

    Daryl Peralta, Joel Casimiro, Aldrin Michael Nilles, Justine Aletta Aguilar, Rowel Atienza, and Rhandley Cajote. Next-best view policy for 3d reconstruction. InEuropean Conference on Computer Vision, pages 558–573. Springer, 2020. 2

    2020

  39. [47]

    A solution to the next best view problem for automated surface acquisition.IEEE Transactions on pattern analysis and machine intelligence, 21(10): 1016–1030, 1999

    Richard Pito. A solution to the next best view problem for automated surface acquisition.IEEE Transactions on pattern analysis and machine intelligence, 21(10): 1016–1030, 1999. 2

    1999

  40. [48]

    A survey on active simultaneous localiza- tion and mapping: State of the art and new frontiers

    Julio A Placed, Jared Strader, Henry Carrillo, Nikolay Atanasov, Vadim Indelman, Luca Carlone, and Jos´e A Castellanos. A survey on active simultaneous localiza- tion and mapping: State of the art and new frontiers. arXiv preprint arXiv:2207.00254, 2022. 2

    2022 arXiv

  41. [49]

    D-nerf: Neural radi- ance fields for dynamic scenes

    Albert Pumarola, Enric Corona, Gerard Pons-Moll, and Francesc Moreno-Noguer. D-nerf: Neural radi- ance fields for dynamic scenes. InProceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 10318–10327, 2021. 1, 2

    2021

  42. [50]

    Occupancy anticipation for efficient ex- ploration and navigation

    Santhosh K Ramakrishnan, Ziad Al-Halah, and Kris- ten Grauman. Occupancy anticipation for efficient ex- ploration and navigation. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, Au- gust 23–28, 2020, Proceedings, Part V 16, pages 400–

    2020

  43. [51]

    Neurar: Neural uncertainty for autonomous 3d recon- struction.arXiv preprint arXiv:2207.10985, 2022

    Yunlong Ran, Jing Zeng, Shibo He, Lincheng Li, Yingfeng Chen, Gimhee Lee, Jiming Chen, and Qi Ye. Neurar: Neural uncertainty for autonomous 3d recon- struction.arXiv preprint arXiv:2207.10985, 2022. 2, 6

    2022 arXiv

  44. [52]

    Habitat: A platform for embodied ai research

    Manolis Savva, Abhishek Kadian, Oleksandr Maksymets, Yili Zhao, Erik Wijmans, Bhavana Jain, Julian Straub, Jia Liu, Vladlen Koltun, Jitendra Malik, et al. Habitat: A platform for embodied ai research. InProceedings of the IEEE/CVF international con- ference on computer vision,...

    2019

  45. [53]

    Vi- sual odometry [tutorial].IEEE robotics & automation magazine, 18(4):80–92, 2011

    Davide Scaramuzza and Friedrich Fraundorfer. Vi- sual odometry [tutorial].IEEE robotics & automation magazine, 18(4):80–92, 2011. 2

    2011

  46. [54]

    Structure-from-motion revisited

    Johannes L Schonberger and Jan-Michael Frahm. Structure-from-motion revisited. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 4104–4113, 2016. 2

    2016

  47. [55]

    Graf: Generative radiance fields for 3d- aware image synthesis.Advances in Neural Informa- tion Processing Systems, 33:20154–20166, 2020

    Katja Schwarz, Yiyi Liao, Michael Niemeyer, and An- dreas Geiger. Graf: Generative radiance fields for 3d- aware image synthesis.Advances in Neural Informa- tion Processing Systems, 33:20154–20166, 2020. 2

    2020

  48. [56]

    A comparison and evaluation of multi-view stereo reconstruction algo- rithms

    Steven M Seitz, Brian Curless, James Diebel, Daniel Scharstein, and Richard Szeliski. A comparison and evaluation of multi-view stereo reconstruction algo- rithms. In2006 IEEE computer society conference on computer vision and pattern recognition (CVPR’06), pages 519–528. IEEE...

    2006

  49. [57]

    MIT press, 2011

    Roland Siegwart, Illah Reza Nourbakhsh, and Da- vide Scaramuzza.Introduction to autonomous mobile robots. MIT press, 2011. 2

    2011

  50. [58]

    Springer, 2009

    Cyrill Stachniss.Robotic mapping and exploration. Springer, 2009. 2

    2009

  51. [59]

    Exploration with active loop-closing for fast- slam

    Cyrill Stachniss, Dirk Hahnel, and Wolfram Bur- gard. Exploration with active loop-closing for fast- slam. In2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)(IEEE Cat. No. 04CH37566), pages 1505–1510. IEEE, 2004. 2

  52. [60]

    The replica dataset: A digital replica of indoor spaces

    Julian Straub, Thomas Whelan, Lingni Ma, Yufan Chen, Erik Wijmans, Simon Green, Jakob J Engel, Raul Mur-Artal, Carl Ren, Shobhit Verma, et al. The replica dataset: A digital replica of indoor spaces. arXiv preprint arXiv:1906.05797, 2019. 6, 7, 1, 2

    1906 arXiv

  53. [61]

    imap: Implicit mapping and positioning in real-time

    Edgar Sucar, Shikun Liu, Joseph Ortiz, and Andrew J Davison. imap: Implicit mapping and positioning in real-time. InProceedings of the IEEE/CVF Interna- tional Conference on Computer Vision, pages 6229– 6238, 2021. 6

    2021

  54. [62]

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

    Cheng Sun, Min Sun, and Hwann-Tzong Chen. Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction. InCVPR, 2022. 1

    2022

  55. [63]

    Stereo matching using belief propagation.IEEE Transactions on pattern analysis and machine intel- ligence, 25(7):787–800, 2003

    Jian Sun, Nan-Ning Zheng, and Heung-Yeung Shum. Stereo matching using belief propagation.IEEE Transactions on pattern analysis and machine intel- ligence, 25(7):787–800, 2003. 2

    2003

  56. [64]

    Probabilistic robotics.Communica- tions of the ACM, 45(3):52–57, 2002

    Sebastian Thrun. Probabilistic robotics.Communica- tions of the ACM, 45(3):52–57, 2002. 2

    2002

  57. [65]

    Active explo- ration in dynamic environments.Advances in neural information processing systems, 4, 1991

    Sebastian B Thrun and Knut M ¨oller. Active explo- ration in dynamic environments.Advances in neural information processing systems, 4, 1991. 2

    1991

  58. [66]

    Edi: Eskf-based disjoint initialization for visual-inertial slam systems

    Weihan Wang, Jiani Li, Yuhang Ming, and Philippos Mordohai. Edi: Eskf-based disjoint initialization for visual-inertial slam systems. In2023 IEEE/RSJ In- ternational Conference on Intelligent Robots and Sys- tems (IROS), pages 1466–1472. IEEE, 2023. 2

    2023

  59. [67]

    Nerf–: Neural radi- ance fields without known camera parameters.arXiv preprint arXiv:2102.07064, 2021

    Zirui Wang, Shangzhe Wu, Weidi Xie, Min Chen, and Victor Adrian Prisacariu. Nerf–: Neural radi- ance fields without known camera parameters.arXiv preprint arXiv:2102.07064, 2021. 2

    2021 arXiv

  60. [68]

    A frontier-based approach for au- tonomous exploration

    Brian Yamauchi. A frontier-based approach for au- tonomous exploration. InProceedings 1997 IEEE In- ternational Symposium on Computational Intelligence in Robotics and Automation CIRA’97. ’Towards New Computational Principles for Robotics and Automa- tion’, pages 146–151. IEEE...

    1997

  61. [69]

    Ac- tive neural mapping

    Zike Yan, Haoxiang Yang, and Hongbin Zha. Ac- tive neural mapping. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 10981–10992, 2023. 4, 6, 2, 3

    2023

  62. [70]

    Mvsnet: Depth inference for unstructured multi-view stereo

    Yao Yao, Zixin Luo, Shiwei Li, Tian Fang, and Long Quan. Mvsnet: Depth inference for unstructured multi-view stereo. InProceedings of the European conference on computer vision (ECCV), pages 767– 783, 2018. 2

    2018

  63. [71]

    pixelnerf: Neural radiance fields from one or few images

    Alex Yu, Vickie Ye, Matthew Tancik, and Angjoo Kanazawa. pixelnerf: Neural radiance fields from one or few images. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 4578–4587, 2021. 1, 2

    2021

  64. [72]

    Mip-splatting: Alias-free 3d gaussian splatting

    Zehao Yu, Anpei Chen, Binbin Huang, Torsten Sat- tler, and Andreas Geiger. Mip-splatting: Alias-free 3d gaussian splatting. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, pages 19447–19456, 2024. 2

    2024

  65. [73]

    Unsupervised learning of monocular depth estimation and visual odometry with deep feature reconstruction

    Huangying Zhan, Ravi Garg, Chamara Saroj Weerasekera, Kejie Li, Harsh Agarwal, and Ian Reid. Unsupervised learning of monocular depth estimation and visual odometry with deep feature reconstruction. InProceedings of the IEEE conference on computer vision and pattern recognitio...

    2018

  66. [74]

    Visual odometry revis- ited: What should be learnt? In2020 IEEE In- ternational Conference on Robotics and Automation (ICRA), pages 4203–4210

    Huangying Zhan, Chamara Saroj Weerasekera, Jia- Wang Bian, and Ian Reid. Visual odometry revis- ited: What should be learnt? In2020 IEEE In- ternational Conference on Robotics and Automation (ICRA), pages 4203–4210. IEEE, 2020. 2

  67. [75]

    Activermap: Radiance field for active mapping and planning.arXiv preprint arXiv:2211.12656, 2022

    Huangying Zhan, Jiyang Zheng, Yi Xu, Ian Reid, and Hamid Rezatofighi. Activermap: Radiance field for active mapping and planning.arXiv preprint arXiv:2211.12656, 2022. 2, 6

    2022 arXiv

  68. [76]

    Nerf++: Analyzing and im- proving neural radiance fields.arXiv preprint arXiv:2010.07492, 2020

    Kai Zhang, Gernot Riegler, Noah Snavely, and Vladlen Koltun. Nerf++: Analyzing and im- proving neural radiance fields.arXiv preprint arXiv:2010.07492, 2020. 1, 2 ActiveGAMER: Active GAussian Mapping through Efficient Rendering Supplementary Material

    2010 arXiv

  69. [77]

    7 delves into additional im- plementation specifics of ActiveGAMER

    Overview In this supplementary material, we provide a detailed out- line structured as follows: Sec. 7 delves into additional im- plementation specifics of ActiveGAMER. Sec. 8 examines the computation costs associated with each module. Com- plementing the results in Sec. 4, Se...

  70. [78]

    Hardware Requirements We conduct our experiments on a desktop PC equipped with a 2.2GHz Intel Xeon E5-2698 CPU and an NVIDIA V100 GPU

    Implementation Details 7.1. Hardware Requirements We conduct our experiments on a desktop PC equipped with a 2.2GHz Intel Xeon E5-2698 CPU and an NVIDIA V100 GPU. Memory consumption varies depending on the scene size. For reference, in an80m 3 scene, the GPU memory and RAM usa...

  71. [79]

    •Gaussian Mapping Module: Updates the Gaussian Map

    Runtime Analysis In this section, we present a runtime analysis of the three major modules in ActiveGAMER: •Data Generation: A simulator generates RGB-D data. •Gaussian Mapping Module: Updates the Gaussian Map. •Rendering-Based Planning Module: Includes a global planner for go...

  72. [80]

    Gdvg”, “gZ6f

    Additional Experimental Results 9.1. Per-scene Results on Replica 3D ReconstructionIn Tab. 5, we present the 3D recon- struction results on the Replica dataset [60], comparing our method with the state-of-the-art active mapping approach, NARUTO [18]. Our method demonstrates co...

Pith tools

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