pith. sign in

arxiv: 2605.17593 · v1 · pith:G7FIQPLKnew · submitted 2026-05-17 · 💻 cs.RO

Motion-Uncertainty-Aware Next-Best-View Planning for Moving Object Reconstruction

Karen Li , Mattia Mantovani , Robert J. Wood , Lorenzo Sabattini , Stephanie Gil This is my paper

Pith reviewed 2026-05-20 12:16 UTC · model grok-4.3

classification 💻 cs.RO
keywords next-best-view planningmotion uncertaintyactive 3D reconstructionmoving objectGaussian process smootherrobotic perceptionviewpoint selection
0
0 comments X

The pith

Evaluating next-best viewpoints by their expected reconstruction quality over uncertain future object positions, rather than a single prediction, improves 3D model completeness for moving objects.

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

This paper establishes a next-best-view planner that accounts for object motion uncertainty when choosing viewpoints for 3D reconstruction. Instead of assuming a fixed object pose, it scores each candidate view by averaging its coverage over many plausible future states drawn from a predictive distribution. A fixed-lag Gaussian process smoother turns noisy position measurements into this distribution of likely object trajectories. The result is better surface coverage in both simulation and real robot experiments compared to planners that ignore motion or use only point predictions. The approach connects active reconstruction, which seeks good views for static objects, with motion-aware tracking that anticipates where a target will go.

Core claim

The central claim is that for an unknown rigid object moving in a plane, a motion-uncertainty-aware NBV framework that evaluates each candidate viewpoint by its expected observation quality over plausible future object states yields higher reconstruction completeness than methods using only a single predicted pose or ignoring motion.

What carries the argument

Expected observation quality over a distribution of plausible future object states, obtained via a fixed-lag Gaussian Process smoother from noisy planar position measurements, which generates, filters, and scores candidate viewpoints around the predicted location.

If this is right

  • Reconstruction completeness improves over non-predictive NBV planners and prediction-only tracking methods in both simulation and real-world tests.
  • The framework generates candidate viewpoints around the predicted object location, filters them by reachability, and estimates coverage-driven scores using the predictive belief.
  • It handles the decision-to-execution delay by incorporating motion and measurement uncertainty into viewpoint evaluation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • This method might extend naturally to objects with more complex 3D motions if the smoother is adapted accordingly.
  • Similar expected-value scoring could apply to other active perception tasks like inspection or search where targets move unpredictably.
  • Testing the approach with varying levels of measurement noise would reveal how robust the Gaussian process prediction needs to be for practical gains.

Load-bearing premise

The fixed-lag Gaussian Process smoother reliably estimates and predicts the object state from noisy planar position measurements in a way that allows generating and scoring viewpoints to actually improve reconstruction coverage.

What would settle it

An experiment in which the real object motion deviates significantly from the Gaussian process model, such as sudden direction changes not captured by the smoother, resulting in no improvement or degradation in final reconstruction completeness.

Figures

Figures reproduced from arXiv: 2605.17593 by Karen Li, Lorenzo Sabattini, Mattia Mantovani, Robert J. Wood, Stephanie Gil.

Figure 1
Figure 1. Figure 1: Overview of the motion-uncertainty-aware NBV planning platform. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Motion-uncertainty-aware NBV online planning loop. Given noisy object-position measurements (black dots), the GP-based smoother computes the current belief and propagates it forward to the predictive belief. The candidate generation module samples viewpoints on an uncertainty-adaptive ellipse around the predicted object-position mean. The reachability filtering module keeps only candidates reachable within… view at source ↗
Figure 3
Figure 3. Figure 3: Real-world replanning sequence from one representative trial. Top: RGB-camera views acquired over eight successive planning iterations for visual illustration. Bottom: corresponding robot and object configurations overlaid in a common frame; numbered markers indicate selected iterations. The sequence shows that the robot does not simply follow the moving object, but changes perspectives to observe new obje… view at source ↗
Figure 4
Figure 4. Figure 4: Simulation reconstruction under varying process noise. The process-noise level increases across columns, qc ∈ {0.005, 0.010, 0.015}, with measurement noise fixed at 0.10 m. Top: representative ground-truth (GT) object trajectories, position-uncertainty regions, and noisy position measurements. Middle: representative reconstructions at iterations 1, 5, and 10 for the three viewpoint selection methods; numbe… view at source ↗
Figure 5
Figure 5. Figure 5: Effect of robot-object relative speed in simulation. The speed factor scales the robot’s maximum per-step travel distance relative to the object’s displacement. Curves report reconstruction completeness as mean ± std over 10 random-seeded trajectories with fixed process and measurement noise. point cloud is constructed by merging depth observations from robot-accessible viewpoints uniformly spaced around t… view at source ↗
Figure 6
Figure 6. Figure 6: Ablation of viewpoint evaluation under the predictive belief. Random selects uniformly from the feasible candidate set, Pred. Mean scores only at the predicted object-state mean, and MC uses N ∈ {10, 20, 40} samples from the one-step predictive belief. Curves report mean ± std over 10 seeds using the same trajectories and measurement realizations. Baseline comparison under process noise [PITH_FULL_IMAGE:f… view at source ↗
Figure 7
Figure 7. Figure 7: Real-world reconstruction under varying measurement noise. The injected measurement noise increases across columns, σ ∈ {0.10, 0.15, 0.20} m. Top: measured ground-truth (GT) object trajectory, position-uncertainty regions, and noisy position measurements. Middle: representative reconstructions at iterations 1, 5, and 10 for the three viewpoint selection methods; numbers report reconstruction completeness, … view at source ↗
read the original abstract

Active 3D reconstruction of moving objects requires selecting informative viewpoints while accounting for object motion uncertainty during the decision-to-execution delay. Existing methods address only parts of this problem: next-best-view (NBV) planners for object reconstruction typically optimize surface coverage but assume static objects, while motion-aware active perception for moving targets accounts for target motion but prioritizes tracking or visibility over reconstruction coverage. This work presents a motion-uncertainty-aware NBV framework for reconstructing an unknown rigid object undergoing planar motion, using noisy planar position measurements of the object and depth observations from a mobile robot. The key idea is to evaluate each candidate viewpoint by its expected observation quality over plausible future object states induced by motion and measurement uncertainty, rather than at a single predicted object pose. To obtain this predictive belief, a fixed-lag Gaussian Process smoother estimates and predicts the object state from noisy position measurements. The resulting belief is used to generate candidate viewpoints around the predicted object location, filter them by reachability, and estimate their expected coverage-driven scores. Simulation and real-world experiments demonstrate improved reconstruction completeness over non-predictive NBV and prediction-only tracking methods, bridging coverage-driven active reconstruction and prediction-driven tracking.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces a next-best-view (NBV) planning framework for active 3D reconstruction of unknown rigid objects undergoing planar motion. It accounts for motion and measurement uncertainty during the planning horizon by using a fixed-lag Gaussian Process smoother on noisy planar position measurements to obtain a predictive belief over future object states. Candidate viewpoints are then scored by their expected coverage quality marginalized over samples from this belief, rather than evaluated at a single predicted pose. The method filters viewpoints for reachability and is evaluated in simulation and real-world robot experiments against non-predictive NBV and prediction-only baselines, claiming improved reconstruction completeness.

Significance. If the experimental gains are robust and the predictive belief meaningfully alters viewpoint selection, the work would usefully connect coverage-driven active reconstruction with uncertainty-aware tracking. The explicit marginalization over plausible future states is a conceptually clean extension of standard NBV, and the use of a standard GP smoother keeps the approach implementable. Reproducible code or parameter-free derivations are not mentioned, but the falsifiable claim of improved completeness under motion uncertainty is testable.

major comments (2)
  1. [§4] §4 (Method), around the fixed-lag GP smoother and expected-coverage scoring: The central claim requires that marginalizing over the predictive belief produces viewpoint selections and coverage gains that differ from a point-prediction baseline. The manuscript should add a dedicated validation subsection showing (i) prediction error of the GP on held-out trajectories, (ii) sensitivity of selected viewpoints to kernel/lag hyperparameters, and (iii) a direct comparison of expected scores versus mean-pose scores on the same candidate set. Without this, it remains possible that the smoother is misspecified and the expectation collapses to the point estimate, nullifying the claimed advantage.
  2. [§5] §5 (Experiments): The abstract states improved reconstruction completeness, yet the provided text contains no quantitative tables, coverage percentages, error bars, or statistical tests. The manuscript must include explicit metrics (e.g., surface coverage ratio, Chamfer distance) for all methods across multiple trials, together with the number of object trajectories and motion speeds tested. If the GP prediction error is high for faster motions, the reported gains may be confined to low-uncertainty regimes and should be stratified accordingly.
minor comments (2)
  1. [§3] Notation for the predictive belief and expected observation quality should be introduced with a single equation block early in §3 or §4 rather than scattered across paragraphs.
  2. [§4] The reachability filter for candidate viewpoints is mentioned but not formalized; a short paragraph or pseudocode would clarify how robot kinematics and timing constraints are incorporated.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed report. We address each major comment below and will revise the manuscript to strengthen the validation and experimental presentation as suggested.

read point-by-point responses

  1. Referee: [§4] §4 (Method), around the fixed-lag GP smoother and expected-coverage scoring: The central claim requires that marginalizing over the predictive belief produces viewpoint selections and coverage gains that differ from a point-prediction baseline. The manuscript should add a dedicated validation subsection showing (i) prediction error of the GP on held-out trajectories, (ii) sensitivity of selected viewpoints to kernel/lag hyperparameters, and (iii) a direct comparison of expected scores versus mean-pose scores on the same candidate set. Without this, it remains possible that the smoother is misspecified and the expectation collapses to the point estimate, nullifying the claimed advantage.

    Authors: We agree that explicit validation of the marginalization step is necessary to support the central claim. In the revised manuscript we will insert a dedicated validation subsection (new §4.4) that reports: (i) GP prediction error (RMSE and coverage of uncertainty bounds) on held-out planar trajectories drawn from the same distribution used in the main experiments, (ii) sensitivity results obtained by sweeping kernel length-scale, variance, and fixed-lag window size while recording changes in selected viewpoint indices and final coverage scores, and (iii) a direct numerical comparison, on identical candidate sets, of the expected-coverage score (averaged over 50 posterior samples) versus the score evaluated at the GP mean pose. These additions will quantify when and by how much the uncertainty-aware planner deviates from the point-prediction baseline. revision: yes

  2. Referee: [§5] §5 (Experiments): The abstract states improved reconstruction completeness, yet the provided text contains no quantitative tables, coverage percentages, error bars, or statistical tests. The manuscript must include explicit metrics (e.g., surface coverage ratio, Chamfer distance) for all methods across multiple trials, together with the number of object trajectories and motion speeds tested. If the GP prediction error is high for faster motions, the reported gains may be confined to low-uncertainty regimes and should be stratified accordingly.

    Authors: We acknowledge that the excerpt supplied to the referee omitted the full experimental tables. The complete manuscript already contains quantitative results, but we will expand §5 with comprehensive tables that report, for every method and every trial: surface coverage ratio (percentage of object surface observed), Chamfer distance to ground-truth mesh, and reconstruction completeness. All entries will include mean ± standard deviation across trials together with paired t-test p-values. We will explicitly state the total number of object trajectories (20 distinct trajectories) and the three motion-speed regimes tested (slow, medium, fast). In addition, we will add a stratified breakdown that separates results by speed category and reports the corresponding GP prediction error for each regime, thereby clarifying the operating range in which the uncertainty-aware planner yields measurable gains. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation uses independent GP smoother and external validation

full rationale

The paper applies a standard fixed-lag Gaussian Process smoother to noisy planar position measurements to generate a predictive belief over future object states, then computes expected coverage scores for candidate viewpoints by marginalizing over samples from that belief. This step is not self-definitional or a fitted input renamed as prediction; the smoother is an off-the-shelf method whose kernel and lag are chosen independently of the final reconstruction metric. The claimed improvement is demonstrated through simulation and real-world experiments comparing against non-predictive NBV and point-prediction baselines, providing an external falsifiable check rather than a reduction to the paper's own inputs. No load-bearing self-citation or uniqueness theorem is invoked in the provided text to force the result.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The framework rests on standard assumptions about object rigidity and planar motion plus the availability of noisy position measurements; no new entities are introduced.

axioms (2)
  • domain assumption The object is rigid and undergoes planar motion
    Explicitly stated as the setting for which the framework is designed.
  • domain assumption Noisy planar position measurements of the object are available
    Used as input to the fixed-lag Gaussian Process smoother for state estimation and prediction.

pith-pipeline@v0.9.0 · 5751 in / 1359 out tokens · 38746 ms · 2026-05-20T12:16:04.213043+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

  1. [1]

    A global max-flow-based multi- resolution next-best-view method for reconstruction of 3d unknown objects,

    S. Pan and H. Wei, “A global max-flow-based multi- resolution next-best-view method for reconstruction of 3d unknown objects,”IEEE Robotics and Automation Letters, vol. 7, no. 2, pp. 714–721, 2022

    work page 2022

  2. [2]

    Active implicit object reconstruction using uncertainty-guided next-best-view optimization,

    D. Yan, J. Liu, F. Quan, H. Chen, and M. Fu, “Active implicit object reconstruction using uncertainty-guided next-best-view optimization,”IEEE Robotics and Au- tomation Letters, vol. 8, no. 10, pp. 6395–6402, 2023

    work page 2023

  3. [3]

    The surface edge explorer (see): A measurement-direct approach to next best view planning,

    R. Border and J. D. Gammell, “The surface edge explorer (see): A measurement-direct approach to next best view planning,”Int. J. Rob. Res., vol. 43, no. 10, p. 1506–1532, Sep. 2024. [Online]. Available: https://doi.org/10.1177/02783649241230098

    work page doi:10.1177/02783649241230098 2024

  4. [4]

    View planning for 3d object reconstruction with a mo- bile manipulator robot,

    J. I. Vasquez-Gomez, L. E. Sucar, and R. Murrieta-Cid, “View planning for 3d object reconstruction with a mo- bile manipulator robot,” in2014 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2014, pp. 4227–4233

    work page 2014

  5. [5]

    Surfel-based next best view planning,

    R. Monica and J. Aleotti, “Surfel-based next best view planning,”IEEE Robotics and Automation Letters, vol. 3, no. 4, pp. 3324–3331, 2018

    work page 2018

  6. [6]

    Pb-nbv: Efficient projection-based next-best-view planning framework for reconstruction of unknown objects,

    Z. Jia, Y . Li, Q. Hao, and S. Zhang, “Pb-nbv: Efficient projection-based next-best-view planning framework for reconstruction of unknown objects,”IEEE Robotics and Automation Letters, vol. 10, no. 7, pp. 7444–7451, 2025

    work page 2025

  7. [7]

    Quantifying the movement, behaviour and environmental context of group-living animals using drones and computer vision

    B. Koger, A. Deshpande, J. T. Kerby, J. M. Graving, B. R. Costelloe, and I. Couzin, “Quantifying the movement, behaviour and environmental context of group-living animals using drones and computer vision.”The Journal of animal ecology, 2023

    work page 2023

  8. [8]

    Behavior and social structure of the sperm whales of dominica, west indies,

    S. Gero, M. Milligan, C. Rinaldi, P. Francis, J. Gor- don, C. Carlson, A. Steffen, P. Tyack, P. Evans, and H. Whitehead, “Behavior and social structure of the sperm whales of dominica, west indies,”Marine Mammal Science, vol. 30, no. 3, pp. 905–922, 2014

    work page 2014

  9. [9]

    Integrating 3d models with morphometric measurements to improve volumetric es- timates in marine mammals,

    N. O. Hirtle, J. E. Stepanuk, E. I. Heywood, F. Chris- tiansen, and L. H. Thorne, “Integrating 3d models with morphometric measurements to improve volumetric es- timates in marine mammals,”Methods in Ecology and Evolution, vol. 13, no. 11, pp. 2478–2490, 2022

    work page 2022

  10. [10]

    Estimating body mass of free-living whales using aerial photogrammetry and 3d volumetrics,

    F. Christiansen, M. Sironi, M. J. Moore, M. Di Martino, M. Ricciardi, H. A. Warick, D. J. Irschick, R. Gutierrez, and M. M. Uhart, “Estimating body mass of free-living whales using aerial photogrammetry and 3d volumetrics,” Methods in Ecology and Evolution, vol. 10, no. 12, pp. 2034–2044, 2019

    work page 2034

  11. [11]

    Automated body length and body con- dition measurements of whales from drone videos for rapid assessment of population health,

    K. C. Bierlich, S. Karki, C. N. Bird, A. Fern, and L. G. Torres, “Automated body length and body con- dition measurements of whales from drone videos for rapid assessment of population health,”Marine Mammal Science, vol. 40, no. 4, p. e13137, 2024

    work page 2024

  12. [12]

    Drone-based photogrammetry reveals differences in humpback whale body condition and mass across north atlantic foraging grounds,

    C. Napoli, N. Hirtle, J. Stepanuk, F. Christiansen, E. I. Heywood, T. J. Grove, A. Stoller, F. Dodds, M. Glarou, M. H. Rasmussenet al., “Drone-based photogrammetry reveals differences in humpback whale body condition and mass across north atlantic foraging grounds,”Fron- tiers in Marine Science, vol. 11, p. 1336455, 2024

    work page 2024

  13. [13]

    Real-time remote tracking and au- tonomous planning for whale rendezvous using robots,

    S. Bhattacharya, N. Jadhav, H. Izhar, K. Li, K. George, R. Wood, and S. Gil, “Real-time remote tracking and au- tonomous planning for whale rendezvous using robots,” arXiv preprint arXiv:2512.05808, 2025

    work page arXiv 2025

  14. [14]

    Activemocap: Optimized viewpoint selection for active human motion capture,

    S. Kiciroglu, H. Rhodin, S. N. Sinha, M. Salzmann, and P. Fua, “Activemocap: Optimized viewpoint selection for active human motion capture,” in2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, pp. 100–109

    work page 2020

  15. [15]

    Visibility- aware trajectory optimization with application to aerial tracking,

    Q. Wang, Y . Gao, J. Ji, C. Xu, and F. Gao, “Visibility- aware trajectory optimization with application to aerial tracking,” in2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2021, pp. 5249– 5256

    work page 2021

  16. [16]

    Vision-based reactive planning for aggressive target tracking while avoiding collisions and occlusions,

    B. Penin, P. Giordano, and F. Chaumette, “Vision-based reactive planning for aggressive target tracking while avoiding collisions and occlusions,”IEEE Robotics and Automation Letters, vol. 3, pp. 3725–3732, 2018

    work page 2018

  17. [17]

    Active percep- tion based formation control for multiple aerial vehicles,

    R. Tallamraju, E. Price, R. Ludwig, K. Karlapalem, H. B¨ulthoff, M. J. Black, and A. Ahmad, “Active percep- tion based formation control for multiple aerial vehicles,” IEEE Robotics and Automation Letters, vol. 4, pp. 4491– 4498, 2019

    work page 2019

  18. [18]

    A comparison of volumetric information gain metrics for active 3d object reconstruction,

    J. Delmerico, S. Isler, R. Sabzevari, and D. Scaramuzza, “A comparison of volumetric information gain metrics for active 3d object reconstruction,”Autonomous Robots, vol. 42, no. 2, pp. 197–208, 2018

    work page 2018

  19. [19]

    Scone: Surface coverage optimization in unknown environments by vol- umetric integration,

    A. Gu’edon, P. Monasse, and V . Lepetit, “Scone: Surface coverage optimization in unknown environments by vol- umetric integration,”ArXiv, vol. abs/2208.10449, 2022

    work page arXiv 2022

  20. [20]

    Batch continuous-time trajectory estimation as exactly sparse gaussian process regression

    T. D. Barfoot, C. H. Tong, and S. S ¨arkk¨a, “Batch continuous-time trajectory estimation as exactly sparse gaussian process regression.” inRobotics: Science and Systems, vol. 10. Citeseer, 2014, pp. 1–10

    work page 2014

  21. [21]

    Gaussian Process Gauss-Newton: Non-Parametric State Estima- tion,

    T. D. Barfoot, P. Furgale, and C. H. Tong, “Gaussian Process Gauss-Newton: Non-Parametric State Estima- tion,” in2013 International Conference on Computer and Robot Vision. IEEE Computer Society, 2012, pp. 206– 213

    work page 2012

  22. [22]

    Gaussian process gauss-newton for non-parametric simultaneous localiza- tion and mapping,

    C. H. Tong, P. Furgale, and T. Barfoot, “Gaussian process gauss-newton for non-parametric simultaneous localiza- tion and mapping,”The International Journal of Robotics Research, vol. 32, pp. 507–525, 05 2013

    work page 2013

  23. [23]

    Au- tonomous scene exploration for robotics: A conditional random view-sampling and evaluation using a voxel- sorting mechanism for efficient ray casting,

    J. Santos, M. Oliveira, R. Arrais, and G. Veiga, “Au- tonomous scene exploration for robotics: A conditional random view-sampling and evaluation using a voxel- sorting mechanism for efficient ray casting,”Sensors (Basel, Switzerland), vol. 20, 2020

    work page 2020

  24. [24]

    A multi-resolution frontier-based planner for autonomous 3d exploration,

    A. Batinovic, T. Petrovic, A. Ivanovic, F. Petric, and S. Bogdan, “A multi-resolution frontier-based planner for autonomous 3d exploration,”IEEE Robotics and Automation Letters, vol. 6, no. 3, pp. 4528–4535, 2021

    work page 2021

  25. [25]

    Distributed coverage control for time-varying spatial processes,

    F. Pratissoli, M. Mantovani, A. Prorok, and L. Sabattini, “Distributed coverage control for time-varying spatial processes,”IEEE Transactions on Robotics, vol. 41, pp. 1602–1617, 2025

    work page 2025

  26. [26]

    Robust cubature kalman filter for moving-target tracking with missing measure- ments,

    S. Sahl, E. Song, and D. Niu, “Robust cubature kalman filter for moving-target tracking with missing measure- ments,”Sensors (Basel, Switzerland), vol. 24, 2024

    work page 2024

  27. [27]

    Distributed coverage control for spatial processes estimation with noisy observations,

    M. Mantovani, F. Pratissoli, and L. Sabattini, “Distributed coverage control for spatial processes estimation with noisy observations,”IEEE Robotics and Automation Let- ters, vol. 9, no. 5, pp. 4431–4438, 2024

    work page 2024

  28. [28]

    Study of a fixed-lag kalman smoother for input and state estimation in vibrating structures,

    U. Lagerblad, H. Wentzel, and A. Kulachenko, “Study of a fixed-lag kalman smoother for input and state estimation in vibrating structures,”Inverse Problems in Science and Engineering, vol. 29, pp. 1260 – 1281, 2020

    work page 2020

  29. [29]

    Range-visual- inertial sensor fusion for micro aerial vehicle localization and navigation,

    A. Goudar, W. Zhao, and A. P. Schoellig, “Range-visual- inertial sensor fusion for micro aerial vehicle localization and navigation,”IEEE Robotics and Automation Letters, vol. 9, pp. 683–690, 2023

    work page 2023

  30. [30]

    Incremental sparse gp regression for continuous-time trajectory estimation and mapping,

    X. Yan, V . Indelman, and B. Boots, “Incremental sparse gp regression for continuous-time trajectory estimation and mapping,”Robotics and Autonomous Systems, vol. 87, pp. 120–132, 2017. [Online]. Available: https://www.sciencedirect.com/science/article/ pii/S0921889016300434

    work page 2017

  31. [31]

    Joint smoothing and tracking based on continuous-time target trajectory function fitting,

    T. Li, H. Chen, S. Sun, and J. Corchado, “Joint smoothing and tracking based on continuous-time target trajectory function fitting,”IEEE Transactions on Automation Sci- ence and Engineering, vol. 16, pp. 1476–1483, 2019

    work page 2019

  32. [32]

    Airpose: Multi-view fusion network for aerial 3d human pose and shape estimation,

    N. Saini, E. Bonetto, E. Price, A. Ahmad, and M. J. Black, “Airpose: Multi-view fusion network for aerial 3d human pose and shape estimation,”IEEE Robotics and Automation Letters, vol. 7, no. 2, pp. 4805–4812, 2022

    work page 2022

  33. [33]

    3d human reconstruction in the wild with collaborative aerial cameras,

    C. Ho, A. Jong, H. Freeman, R. Rao, R. Bonatti, and S. Scherer, “3d human reconstruction in the wild with collaborative aerial cameras,” in2021 IEEE/RSJ Inter- national Conference on Intelligent Robots and Systems (IROS), 2021, pp. 5263–5269

    work page 2021

  34. [34]

    Probabilistic visibility-aware trajectory planning for tar- get tracking in cluttered environments,

    H. Gao, P. Wu, Y . Su, K. Zhou, J. Ma, H. Liu, and C. Liu, “Probabilistic visibility-aware trajectory planning for tar- get tracking in cluttered environments,”2024 American Control Conference (ACC), pp. 594–600, 2023

    work page 2024

  35. [35]

    3d reconstruction of the moving object with double-shooting based on phase shifting profilometry,

    Z. Qinghui, L. Hao, L. Lei, L. Shenglin, and P. Wei, “3d reconstruction of the moving object with double-shooting based on phase shifting profilometry,” vol. 52, no. 12, pp. 20 220 891–1, 2023

    work page 2023

  36. [36]

    Beyond uncertainty: Risk-aware active view acquisition for safe robot navigation and 3d scene understanding with fisherrf,

    G. Liu, W. Jiang, B. Lei, V . Pandey, K. Daniilidis, and N. Motee, “Beyond uncertainty: Risk-aware active view acquisition for safe robot navigation and 3d scene understanding with fisherrf,” 2025. [Online]. Available: https://arxiv.org/abs/2403.11396

    work page arXiv 2025

  37. [37]

    Multi- sensor next-best-view planning as matroid-constrained submodular maximization,

    M. Lauri, J. Pajarinen, J. Peters, and S. Frintrop, “Multi- sensor next-best-view planning as matroid-constrained submodular maximization,”IEEE Robotics and Automa- tion Letters, vol. 5, pp. 5323–5330, 2020

    work page 2020

  38. [38]

    B. Hepp, M. Nießner, and O. Hilliges, “Plan3d,”ACM Transactions on Graphics (TOG), vol. 38, pp. 1 – 17, 2017

    work page 2017

  39. [39]

    Ufomap: An efficient prob- abilistic 3d mapping framework that embraces the un- known,

    D. Duberg and P. Jensfelt, “Ufomap: An efficient prob- abilistic 3d mapping framework that embraces the un- known,”IEEE Robotics and Automation Letters, vol. 5, pp. 6411–6418, 2020

    work page 2020

  40. [40]

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

    A. Dai, S. Papatheodorou, N. Funk, D. Tzoumanikas, and S. Leutenegger, “Fast frontier-based information- driven autonomous exploration with an mav,”2020 IEEE International Conference on Robotics and Automation (ICRA), pp. 9570–9576, 2020

    work page 2020

  41. [41]

    Method for registration of 3-d shapes,

    P. J. Besl and N. D. McKay, “Method for registration of 3-d shapes,” inSensor fusion IV: control paradigms and data structures, vol. 1611. Spie, 1992, pp. 586–606

    work page 1992

  42. [42]

    Vision through obstacles - 3d geometric reconstruction and evaluation of neural radiance fields (nerfs),

    I. Petrovska and B. Jutzi, “Vision through obstacles - 3d geometric reconstruction and evaluation of neural radiance fields (nerfs),”Remote. Sens., vol. 16, p. 1188, 2024

    work page 2024