Pith. sign in

REVIEW 4 major objections 5 minor 66 references

A shared signed-distance map can drive both mapping and planning, letting a quadrotor navigate unseen indoor spaces in real time.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 12:47 UTC pith:ZFKMZOYM

load-bearing objection The integrated OREN-Bubble* system is real and the flight demo is genuine, but the Bubble* pseudocode as printed cannot expand the first bubble, so the completeness proof rests on an unstated fix; still worth refereeing. the 4 major comments →

arxiv 2607.19306 v1 pith:ZFKMZOYM submitted 2026-07-21 cs.RO cs.AIcs.CVcs.SYeess.SY

From Distances to Trajectories: Real-Time Signed Distance Function Mapping and Distance-Accelerated Motion Planning for UAVs

Jason Stanley (1) , Zhirui Dai (1) , Qihao Qian (1) , Tzu-Chin Ho (1) , Tianxing Fan (1) , Siddharth Saha (2) , Christopher Barngrover (2) , Ki Myung Brian Lee (1)
show 6 more authors
Nikolay Atanasov (1) ((1) UC San Diego La Jolla USA (2) Shield AI San Diego USA)
This is my paper
classification cs.RO cs.AIcs.CVcs.SYeess.SY
keywords signed distance functionoctree residual networkimplicit neural representationmotion planningbubble corridorsearch-based planningquadrotor autonomyreal-time mapping
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper tries to establish that mapping and motion planning for autonomous flight belong in a single representation: a signed distance function (SDF), which returns the clearance to the nearest obstacle at every point. It introduces OREN, an online SDF reconstructor that combines a coarse explicit octree prior with a compact neural residual, and Bubble*, a search-based planner that treats each SDF query as a maximal collision-free ball and searches a graph of such balls. Together they let a quadrotor build a map and plan trajectories of about 90 meters in 1–3 seconds, issuing 91–99% fewer collision checks than a grid-based A* search. If the paper is right, distance information—not just occupancy—is the natural currency for both the map and the planner.

Core claim

The central claim is that an accurate, non-truncated, differentiable SDF can be reconstructed online and then used as the sole geometric model for planning. OREN stores learnable SDF values and gradients at octree vertices, obtains a coarse prior by gradient-augmented interpolation, and decodes a neural residual from implicit features to recover fine surface detail; an occupancy decoder supervises the sign of the prediction. Bubble* queries this SDF at grid nodes, grows an open ball of radius equal to the signed distance at each node, and expands a graph whose edges connect nodes inside a bubble to nodes on its boundary, so one distance query validates an entire region. The paper proves term

What carries the argument

The bubble is the central object: for a grid node c, the SDF value d(c) defines an open ball B(c,d(c)) of guaranteed-free space. Bubble* expands these bubbles by adding boundary nodes—nodes with a grid neighbor outside the ball—as successors, so a single distance query replaces many per-cell collision checks and the search can jump across large free regions. OREN supplies the SDF: a semi-sparse octree (shallower layers kept dense so interpolation stays accurate away from surfaces) stores learnable distance and gradient values at vertices, gradient-augmented trilinear interpolation forms a coarse prior, and a small MLP decodes implicit vertex features into a residual correction, producing a c

Load-bearing premise

The completeness guarantee only holds when a 'clear grid path' exists—a grid path where every node has signed distance at least 1.5 times the grid resolution (Definition 1, Theorem 2); in unknown or narrow environments such a path may not exist even when a safe trajectory does, and the paper does not quantify how often this happens or provide a fallback.

What would settle it

In a grid of resolution Δ, build a straight corridor of width 2.0Δ (so centerline clearance is 1.0Δ, below 1.5Δ) and run Bubble* end-to-end; if it reports failure despite a centered straight trajectory being collision-free, the completeness precondition is shown to exclude a navigable environment.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • A planner that consumes signed distance directly can replace per-cell collision checks with one distance query per bubble, reducing collision checks by 91–99% in the paper's experiments.
  • Because the bubble corridor is produced by the search itself, path-finding and safe-corridor construction cease to be separate stages, which keeps trajectory optimization fast (e.g., 523 ms vs 1005 ms in the industrial test).
  • Non-truncated SDF support means distance queries far from surfaces remain valid, which is what lets global planning reason about clearance across large spaces rather than only near surfaces.
  • The termination, completeness, and failure-detection guarantees mean the planner can act as a decision module: it returns a corridor when one with sufficient clearance exists and reports failure otherwise.
  • If OREN's mapping accuracy transfers, the same loop can run on compute-limited platforms, since the full mapping-planning stack ran onboard in real time in the paper's flight tests.

Where Pith is reading between the lines

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

  • A testable extension is to lower the 1.5Δ clearance precondition: in a corridor whose width is between 2Δ and 3Δ, a centerline path may be safe for a small vehicle yet fail the clear-grid-path condition, so Bubble* would report failure; a resolution-adaptive or hybrid fallback would close this gap.
  • The planner only needs a distance oracle, so the same bubble-graph idea could be applied to configuration-space distance fields for manipulators or to learned distance fields from other sensors, not just OREN.
  • The safety margin r in the trajectory optimization depends on the accuracy of far-field SDF values; if OREN's error grows with distance in unseen areas, the effective clearance shrinks, suggesting that an online uncertainty estimate for bubbles would strengthen the guarantees.
  • Because multiple Bubble* plans can be concatenated into one overlapping-bubble sequence, the method supports waypoint-chained missions without re-planning from scratch, which the paper demonstrates with a looping three-segment trajectory.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes an integrated mapping and planning system for UAVs built around a non-truncated signed distance function (SDF). The mapping component, OREN, combines an explicit octree prior with an implicit neural residual to reconstruct a differentiable SDF online from depth point clouds. The planning component, Bubble*, searches over a graph of collision-free 'bubbles' grown from SDF clearance values, then uses MINCO to optimize a dynamically feasible trajectory within the resulting bubble corridor. The authors claim that OREN improves SDF estimation by 22% over baselines, that Bubble* finds trajectories spanning ~90 m in 1–3 s versus up to 10 s for baselines, and that the full pipeline runs onboard in real time on a Jetson Orin NX. Formal termination, completeness, and failure-detection guarantees are stated for Bubble*, and experiments include 2D and 3D simulated comparisons plus a real-world quadrotor flight.

Significance. If the central claims hold, this work makes a convincing case that mapping and planning should be co-designed around one distance representation: OREN provides the continuous clearance information that Bubble* consumes directly, and the planner's bubble graph unifies path search and safe-corridor construction. The formal completeness/failure-detection analysis, the code availability, and the real onboard demonstration are notable strengths. However, the evidence presented is thin in key places: the 3D planning results are single runs with no error bars, the 22% mapping improvement is inherited from the authors' prior conference paper rather than freshly evaluated here, and the printed search algorithm contains a load-bearing bug that prevents it from running as described. These issues do not invalidate the overall approach, but they must be addressed before the claims can be fully trusted.

major comments (4)
  1. [Sec. 5.2, Algorithm 1, line 20] The algorithm as printed cannot perform its first bubble expansion. On initialization OPEN={p_s}; the first iteration pops c=p_s and removes it from OPEN (line 6) before constructing B_c and calling CalculateSuccessors. Line 20 then takes argmin over k in OPEN∩B_c, which is empty on the first call. No successor receives a finite parent, no node is inserted into OPEN, and the loop exits on line 17 with failure. Thus the printed pseudocode returns failure for every nontrivial start–goal pair. The prose in Sec. 5.2 says the predecessor is selected 'among nodes inside B_c', which differs from the coded rule, and the completeness proof in Sec. 5.3 assumes each boundary node is inserted into OPEN with a finite parent cost, which cannot happen as written. This is load-bearing for termination, completeness, and failure detection. Please correct the pseudocode to match the intended rule — e.g., c
  2. [Table 4] The 3D planning comparison reports a single run per planner per environment, with no variance, seeds, or confidence intervals. The central efficiency claim — Bubble* has the lowest total planning time while matching path quality — rests on one trajectory. This is especially problematic for the stochastic RRT/RRT* baselines, where run-to-run variation can easily exceed the observed margins (e.g., Industrial: Bubble* 1008 ms vs A* 1228 ms; Forest: Bubble* 3776 ms vs A* 4152 ms). Please report multiple independent runs with standard deviations or interquartile ranges, and state the number of trials. Without this, the 3D planning conclusions are not quantitatively supported.
  3. [Sec. 6.1 and abstract] The headline '22% SDF improvement' is not a fresh result of this manuscript. Table 2 is explicitly a summary of the authors' prior conference paper (Dai, Qian, et al., 2026), and the abstract and introduction present the 22% number as if it were newly demonstrated here. Since the current paper's contribution is the integrated OREN–Bubble* system, the mapping accuracy claim should either be re-evaluated in the paper's own simulated environments or clearly attributed as prior work in every appearance (abstract, introduction, and conclusion). As written, the 22% claim is effectively self-cited and does not provide independent evidence for the current system.
  4. [Sec. 5.3, Definition 1 and Theorem 2] The completeness guarantee is conditional on the existence of a clear grid path with clearance at least 1.5Δ, and the text states that 1.5Δ is 'the smallest' clearance admissible under the grid-occupancy model without derivation. Mathematically, d(p_j)>Δ already ensures that the open bubble B(p_j,d(p_j)) contains the axis-aligned grid neighbors at distance Δ; the additional factor 1.5 is not justified in the proof. Furthermore, in unknown or partially mapped environments such a 1.5Δ-clear path may not exist even when a safe (nonzero-clearance) trajectory does, and the paper does not quantify how often Bubble* would report failure in navigable situations, nor provide a fallback. Please substantiate the 1.5Δ choice or revise the statement, and discuss the practical implications of this completeness condition.
minor comments (5)
  1. [Throughout] Typographical and formatting issues: the title and abstract use 'UA Vs' with a space; Algorithm 1 uses the unusual arrow '←−' instead of '←'; the GitHub link in Data Availability contains a space ('erl oren bubble star demo') and should be a single URL.
  2. [Sec. 6.2 vs Sec. 6.4] The simulated update-rate results (Fig. 6) report OREN running at 19.1–28.4 Hz, while the real-world experiment states OREN runs at 7 Hz. Please clarify whether this discrepancy is due to different hardware, point-cloud preprocessing, or parameter settings, and how the 30 Hz depth stream is handled at 7 Hz mapping rate.
  3. [Table 3] The text says OREN achieves 'better or comparable' occupancy prediction, but in the Industrial environment OctoMap has a higher F1 (0.710 vs 0.688). This is not a major issue, but the claim should be phrased to reflect the actual pattern (higher recall, lower precision) rather than a blanket superiority.
  4. [Sec. 5.4] The disk-overlap construction in Eq. (14) is presented for two arbitrary overlapping balls, but the relationship to the specific bubble sequence returned by Bubble* (where consecutive bubble centers lie inside the previous bubble) is not formally connected. Also, the notation q_i is overloaded: Eq. (16) uses q_i for waypoints, while earlier q denotes a general point. Please align notation and make the overlap-construction validity conditions explicit.
  5. [Sec. 2.2] The related-work discussion of bubble-based planning mentions Ren et al. (2022) and the authors' prior work K.M.B. Lee et al. (2024), but the latter is not listed in the experimental comparisons. Given that Prior Bubble Cover is the closest algorithmic predecessor, a sentence explaining what Bubble* adds beyond it beyond the citation would help position the contribution.

Circularity Check

1 steps flagged

No equation-level circularity; the main circularity concern is that OREN's headline 22% improvement is imported from the authors' own prior paper, while Bubble*'s formal derivation is independent (though the printed Algorithm 1 has a serious non-circular correctness bug).

specific steps
  1. self citation load bearing [Abstract; Sec. 6.1 and Table 2 ('Our prior work (Dai, Qian, et al., 2026) compares OREN extensively against SDF mapping baselines, and Sec. 6.1 summarizes those results.')]
    "OREN improves SDF estimation by 22% compared to baselines... Our prior work (Dai, Qian, et al., 2026) compares OREN extensively against SDF mapping baselines, and Sec. 6.1 summarizes those results."

    The 22% SDF-improvement figure is the paper's headline quantitative evidence for OREN, but Table 2 is explicitly a summary of the authors' own IROS paper, not a result derived or independently reproduced in this manuscript. The same-author citation is the only support connecting the OREN variant in this paper to that number. This is self-citation doing the load-bearing work: the claimed accuracy advantage is imported from the authors' prior benchmark rather than obtained from a first-principles relation or a fresh independent evaluation, so it cannot independently confirm the current paper's central mapping claim.

full rationale

Aside from that empirical self-citation, the derivation chain is self-contained. Bubble*'s Theorem 2 is conditional on Definition 1 (a clear grid path with d >= 1.5 Delta); the induction uses only the geometry of open balls and grid adjacency, and the trajectory optimizer is a standard MINCO formulation. No parameter is fitted to the reported trajectories and then renamed a prediction; OREN's SDF is consumed by Bubble* as an oracle rather than being chosen to make the planner succeed. The printed Algorithm 1 has a serious correctness bug (line 20 selects parents from OPEN intersect B_c after line 6 removes c, which can leave OPEN empty on the first expansion), and Theorem 1's appeal to standard A* optimality is questionable for the printed update rule; these are correctness defects, not circular reductions, so they are excluded from the circularity score. The unsubstantiated '1.5 Delta is the smallest' statement is likewise an unsupported assertion, not circularity. Overall, the central formal derivation has independent content (for a corrected pseudocode), and the empirical claims are externally testable, so the modest score comes from the self-cited 22% claim.

Axiom & Free-Parameter Ledger

8 free parameters · 6 axioms · 0 invented entities

The central claims depend on several hand-chosen hyperparameters (octree resolution, feature dimension, MLP size, safety radius) and on the clearance-path assumption for the completeness proof. The mapping architecture itself is inherited from prior work, so many design choices are not re-justified in this paper.

free parameters (8)
  • octree resolution ℓ = 10 cm
    Controls prior accuracy vs memory; configuration from prior work (Dai, Qian, et al., 2026).
  • octree layers N and semi-sparse layers M = N=8, M=5
    Chosen in prior work; affects memory and interpolation accuracy.
  • implicit feature dimension F = 3
    Hand-chosen; higher dimensions could improve residual accuracy but cost memory.
  • MLP hidden size = 32
    Two 32-dimensional hidden layers with LeakyReLU; not optimized.
  • occupancy confidence margin τ = 3
    Used in sign-consistency loss; hand-set.
  • planning grid resolution Δ = not reported
    Grid discretization for Bubble*; the value used in experiments is not stated.
  • safety radius r = not reported
    Robot clearance in trajectory optimization (Eq. 4); not specified.
  • clearance threshold 1.5Δ = 1.5Δ
    Used in the completeness proof; claimed to be the smallest admissible clearance under the grid-occupancy model.
axioms (6)
  • standard math SDF satisfies the Eikonal equation and unit gradient a.e.
    Used to justify gradient-augmented interpolation and the differentiability of the representation (Eq. 2).
  • domain assumption Quadrotor dynamics are differentially flat with flat output (p, ψ)
    Standard result (Mellinger & Kumar, 2011) used to reduce planning to a position trajectory.
  • standard math Euclidean distance heuristic h(p)=||p-p_g|| is consistent
    Used in Bubble* termination proof; holds by triangle inequality.
  • domain assumption Existence of a clear grid path with clearance ≥ 1.5Δ
    Completeness is conditional on this; may not hold in cluttered scenes.
  • domain assumption Ray-cast free/occupied labels are correct
    OREN occupancy supervision relies on ray-casting labels from depth measurements.
  • domain assumption MINCO trajectory class can express the optimal solution
    Trajectory optimization is restricted to the MINCO polynomial parameterization.

pith-pipeline@v1.3.0-alltime-deepseek · 23994 in / 16337 out tokens · 153219 ms · 2026-08-01T12:47:59.094859+00:00 · methodology

0 comments
read the original abstract

Autonomous flight in cluttered environments requires a robot to build a geometric map of its surroundings and plan safe, dynamically feasible trajectories, all onboard and in real time. Conventional approaches treat mapping and planning as separate stages and often rely on binary occupancy for collision checking. We argue that these two stages should be co-designed around a single representation: a signed distance function (SDF). By encoding distance to the nearest obstacle, an SDF provides richer information for planning and trajectory optimization than occupancy alone. We develop an Octree REsidual Network (OREN) that pairs an explicit octree prior with an implicit neural residual to reconstruct SDFs online from point cloud observations with the efficiency of volumetric methods and the accuracy and differentiability of neural methods. In tandem, we develop Bubble$^\star$, a search-based planner that exploits the distance information to grow maximal collision-free balls, which we call bubbles, with formal guarantees of termination, completeness, and failure detection. Planning over a graph of bubbles significantly reduces collision checks compared to a grid-based A$^\star$ search and returns a bubble sequence that forms a safe corridor for trajectory optimization. We demonstrate the integrated OREN-Bubble$^\star$ approach onboard a quadrotor, navigating unseen indoor environments in real time under tight compute constraints. OREN improves SDF estimation by $22$% compared to baselines, while Bubble$^\star$ finds trajectories spanning $\approx 90$ m through a cluttered environment in $1$-$3$ sec., whereas baselines take up to $10$ sec. in the same environment.

discussion (0)

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

Reference graph

Works this paper leans on

66 extracted references · 5 canonical work pages

  1. [1]

    , Chakravorty, S

    agha2011firm APACrefauthors Agha-Mohammadi, A A. , Chakravorty, S. Amato, N.M. APACrefauthors \ 2011 . FIRM : Feedback controller-based Information-state RoadMap-a framework for motion planning under uncertainty FIRM : Feedback controller-based information-state roadmap-a framework for motion planning under uncertainty . 2011 IEEE/RSJ International Confer...

    2011

  2. [2]

    \ Pavone, M

    allen2019realtime APACrefauthors Allen, R.E. \ Pavone, M. APACrefauthors \ 2019 . A real-time framework for kinodynamic planning in dynamic environments with application to quadrotor obstacle avoidance A real-time framework for kinodynamic planning in dynamic environments with application to quadrotor obstacle avoidance . Robotics and Autonomous Systems 1...

    work page doi:10.1016/j.robot.2018.11.017 2019

  3. [3]

    , Rizzi, G

    brunner_aerial_manipulation APACrefauthors Brunner, M. , Rizzi, G. , Studiger, M. , Siegwart, R. Tognon, M. APACrefauthors \ 2022 . A Planning-and-Control Framework for Aerial Manipulation of Articulated Objects A planning-and-control framework for aerial manipulation of articulated objects . IEEE Robotics and Automation Letters 7 4 10689-10696, APACrefDO...

    arXiv 2022

  4. [4]

    \ Levoy, M

    curless_volumetric_1996 APACrefauthors Curless, B. \ Levoy, M. APACrefauthors \ 1996 . A Volumetric Method for Building Complex Models from Range Images A volumetric method for building complex models from range images . Conference on Computer Graphics and Interactive Techniques (SIGGRAPH) Conference on Computer Graphics and Interactive Techniques (SIGGRA...

    1996

  5. [5]

    , Fan, T

    dai2026kernelsdf APACrefauthors Dai, Z. , Fan, T. , Amani, M. , Seo, J. , Lee, K.M.B. , Oh, H. Atanasov, N. APACrefauthors \ 2026 . Kernel-SDF : An Open-Source Library for Real-Time Signed Distance Function Estimation Using Kernel Regression. Kernel-SDF : An open-source library for real-time signed distance function estimation using kernel regression. Pre...

    Pith/arXiv arXiv 2026

  6. [6]

    , Qian, Q

    dai_oren_2025 APACrefauthors Dai, Z. , Qian, Q. , Fan, T. Atanasov, N. APACrefauthors \ 2026 . OREN : Octree Residual Network for Real-Time E uclidean Signed Distance Mapping OREN : Octree Residual Network for Real-Time E uclidean Signed Distance Mapping . IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE/RSJ International ...

    2026

  7. [7]

    \ Tedrake, R

    Deits2015 APACrefauthors Deits, R. \ Tedrake, R. APACrefauthors \ 2015 . Computing Large Convex Regions of Obstacle-Free Space Through Semidefinite Programming Computing large convex regions of obstacle-free space through semidefinite programming . H.L. Akin, N.M. Amato, V. Isler \ A.F. van der Stappen\ ( ), Algorithmic Foundations of Robotics XI: Selecte...

    2015

  8. [8]

    , Dang, T

    dharmadhikari2020motionprimitives APACrefauthors Dharmadhikari, M. , Dang, T. , Solanka, L. , Loje, J. , Nguyen, H. , Khedekar, N. Alexis, K. APACrefauthors \ 2020 . Motion Primitives-based Path Planning for Fast and Agile Exploration using Aerial Robots Motion primitives-based path planning for fast and agile exploration using aerial robots . IEEE Intern...

    2020

  9. [9]

    , Franchi, A

    faessler2018differential APACrefauthors Faessler, M. , Franchi, A. Scaramuzza, D. APACrefauthors \ 2018 04 . Differential Flatness of Quadrotor Dynamics Subject to Rotor Drag for Accurate Tracking of High-Speed Trajectories Differential flatness of quadrotor dynamics subject to rotor drag for accurate tracking of high-speed trajectories . IEEE Robotics an...

    arXiv 2018

  10. [10]

    , Tarrio, J

    funk2021multires APACrefauthors Funk, N. , Tarrio, J. , Papatheodorou, S. , Popovi\' c , M. , Alcantarilla, P.F. Leutenegger, S. APACrefauthors \ 2021 April . Multi-Resolution 3D Mapping with Explicit Free Space Representation for Fast and Accurate Mobile Robot Motion Planning Multi-resolution 3D mapping with explicit free space representation for fast an...

    arXiv 2021

  11. [11]

    gao2019flying APACrefauthors Gao, F. , Wu, W. , Gao, W. Shen, S. APACrefauthors \ 2019 . Flying on point clouds: Online trajectory generation and autonomous navigation for quadrotors in cluttered environments Flying on point clouds: Online trajectory generation and autonomous navigation for quadrotors in cluttered environments . Journal of Field Robotics ...

    doi:10.1002/rob.21842 2019

  12. [12]

    , Yariv, L

    gropp_implicit_2020 APACrefauthors Gropp, A. , Yariv, L. , Haim, N. , Atzmon, M. Lipman, Y. APACrefauthors \ 2020 . Implicit Geometric Regularization for Learning Shapes Implicit Geometric Regularization for Learning Shapes . International Conference on Machine Learning . International Conference on Machine Learning . JMLR

    2020

  13. [13]

    , Gao, F

    han_fiesta_2019 APACrefauthors Han, L. , Gao, F. , Zhou, B. Shen, S. APACrefauthors \ 2019 . FIESTA : Fast Incremental E uclidean Distance Fields for Online Motion Planning of Aerial Robots FIESTA : Fast incremental E uclidean distance fields for online motion planning of aerial robots . IEEE/RSJ International Conference on Intelligent Robots and Systems ...

    2019

  14. [14]

    \ Grastien, A

    harabor2011jumppointsearch APACrefauthors Harabor, D. \ Grastien, A. APACrefauthors \ 2011 . Online Graph Pruning for Pathfinding on Grid Maps Online graph pruning for pathfinding on grid maps . Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence (AAAI) Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence (AAAI...

    2011

  15. [15]

    , Nilsson, N.J

    heuristic_basis APACrefauthors Hart, P.E. , Nilsson, N.J. Raphael, B. APACrefauthors \ 1968 . A Formal Basis for the Heuristic Determination of Minimum Cost Paths A formal basis for the heuristic determination of minimum cost paths . IEEE Transactions on Systems Science and Cybernetics 4 2 100-107, APACrefDOI doi:10.1109/TSSC.1968.300136 APACrefDOI

    arXiv 1968

  16. [16]

    , Wurm, K.M

    hornung13auro APACrefauthors Hornung, A. , Wurm, K.M. , Bennewitz, M. , Stachniss, C. Burgard, W. APACrefauthors \ 2013 . OctoMap : An Efficient Probabilistic 3D Mapping Framework Based on Octrees OctoMap : An efficient probabilistic 3D mapping framework based on octrees . Autonomous Robots 34 3 189-206, APACrefDOI doi:10.1007/s10514-012-9321-0 APACrefDOI...

    doi:10.1007/s10514-012-9321-0 2013

  17. [17]

    , Harms, M

    jacquet2025neural APACrefauthors Jacquet, M. , Harms, M. Alexis, K. APACrefauthors \ 2025 . Neural NMPC through signed distance field encoding for collision avoidance Neural NMPC through signed distance field encoding for collision avoidance . The International Journal of Robotics Research , APACrefDOI doi:10.1177/02783649251401223 APACrefDOI OnlineFirst,...

    doi:10.1177/02783649251401223 2025

  18. [18]

    , Wang, Z

    mapless_planner APACrefauthors Ji, J. , Wang, Z. , Wang, Y. , Xu, C. Gao, F. APACrefauthors \ 2021 . Mapless-Planner: A Robust and Fast Planning Framework for Aggressive Autonomous Flight without Map Fusion Mapless-planner: A robust and fast planning framework for aggressive autonomous flight without map fusion . IEEE International Conference on Robotics ...

    2021

  19. [19]

    , Zhang, H

    jiang_h2-mapping_2023 APACrefauthors Jiang, C. , Zhang, H. , Liu, P. , Yu, Z. , Cheng, H. , Zhou, B. Shen, S. APACrefauthors \ 2023 . H2-Mapping : Real-Time Dense Mapping Using Hierarchical Hybrid Representation H2-Mapping : Real-time dense mapping using hierarchical hybrid representation . IEEE Robotics and Automation Letters 8 10 6787-6794, APACrefDOI d...

    arXiv 2023

  20. [20]

    \ Frazzoli, E

    karaman_sampling APACrefauthors Karaman, S. \ Frazzoli, E. APACrefauthors \ 2011 . Sampling-Based Algorithms for Optimal Motion Planning Sampling-based algorithms for optimal motion planning . The International Journal of Robotics Research 30 7 846-894, APACrefDOI doi:10.1177/0278364911406761 APACrefDOI

    doi:10.1177/0278364911406761 2011

  21. [21]

    , Adrian Prisacariu, V

    kahler_very_2015 APACrefauthors Kähler, O. , Adrian Prisacariu, V. , Yuheng Ren, C. , Sun, X. , Torr, P. Murray, D. APACrefauthors \ 2015 . Very High Frame Rate Volumetric Integration of Depth Images on Mobile Devices Very high frame rate volumetric integration of depth images on mobile devices . IEEE Transactions on Visualization and Computer Graphics (T...

    arXiv 2015

  22. [22]

    , Boche, S

    laina2025scalable APACrefauthors Laina, S.B. , Boche, S. , Papatheodorou, S. , Tzoumanikas, D. , Schaefer, S. , Chen, H. Leutenegger, S. APACrefauthors \ 2025 . Scalable Outdoors Autonomous Drone Flight with Visual-Inertial SLAM and Dense Submaps Built without LiDAR Scalable outdoors autonomous drone flight with visual-inertial SLAM and dense submaps buil...

    2025

  23. [23]

    \ Kuffner, J.J

    lavalle2001rrt APACrefauthors LaValle, S.M. \ Kuffner, J.J. APACrefauthors \ 2001 . Randomized Kinodynamic Planning Randomized kinodynamic planning . The International Journal of Robotics Research 20 5 378--400, APACrefDOI doi:10.1177/02783640122067453 APACrefDOI

    doi:10.1177/02783640122067453 2001

  24. [24]

    , Zhang, C

    lee2019gpis APACrefauthors Lee, B. , Zhang, C. , Huang, Z. Lee, D.D. APACrefauthors \ 2019 . Online Continuous Mapping Using G aussian Process Implicit Surfaces Online continuous mapping using G aussian process implicit surfaces . IEEE International Conference on Robotics and Automation (ICRA) IEEE International Conference on Robotics and Automation (ICRA...

    2019

  25. [25]

    , Dai, Z

    lee2024safebubblecovermotion APACrefauthors Lee, K.M.B. , Dai, Z. , Le Gentil, C. , Wu, L. , Atanasov, N. Vidal-Calleja, T. APACrefauthors \ 2024 . Safe Bubble Cover for Motion Planning on Distance Fields. Safe bubble cover for motion planning on distance fields. Preprint at https://arxiv.org/abs/2408.13377

    Pith/arXiv arXiv 2024

  26. [26]

    , Chi, X

    li2024config APACrefauthors Li, Y. , Chi, X. , Razmjoo, A. Calinon, S. APACrefauthors \ 2024 . Configuration Space Distance Fields for Manipulation Planning Configuration space distance fields for manipulation planning . Robotics: Science and Systems (RSS) . Robotics: Science and Systems (RSS)

    2024

  27. [27]

    , Zhang, Y

    li2024robot APACrefauthors Li, Y. , Zhang, Y. , Razmjoo, A. Calinon, S. APACrefauthors \ 2024 . Representing Robot Geometry as Distance Fields: Applications to Whole-Body Manipulation Representing robot geometry as distance fields: Applications to whole-body manipulation . IEEE International Conference on Robotics and Automation (ICRA) IEEE International ...

    2024

  28. [28]

    \ Ferguson, D

    likhachev2009lattice APACrefauthors Likhachev, M. \ Ferguson, D. APACrefauthors \ 2009 . Planning Long Dynamically Feasible Maneuvers for Autonomous Vehicles Planning long dynamically feasible maneuvers for autonomous vehicles . The International Journal of Robotics Research 28 8 933--945, APACrefDOI doi:10.1177/0278364909340445 APACrefDOI

    work page doi:10.1177/0278364909340445 2009

  29. [29]

    , Zhang, K

    ReDSDF APACrefauthors Liu, P. , Zhang, K. , Tateo, D. , Jauhri, S. , Peters, J. Chalvatzaki, G. APACrefauthors \ 2022 . Regularized Deep Signed Distance Fields for Reactive Motion Generation Regularized deep signed distance fields for reactive motion generation . IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) IEEE/RSJ Internati...

    2022

  30. [30]

    , Atanasov, N

    liu2017search APACrefauthors Liu, S. , Atanasov, N. , Mohta, K. Kumar, V. APACrefauthors \ 2017 . Search-based motion planning for quadrotors using linear quadratic minimum time control Search-based motion planning for quadrotors using linear quadratic minimum time control . IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) IEEE/R...

    2017

  31. [31]

    , Watterson, M

    liu_corridor APACrefauthors Liu, S. , Watterson, M. , Mohta, K. , Sun, K. , Bhattacharya, S. , Taylor, C.J. Kumar, V. APACrefauthors \ 2017 . Planning Dynamically Feasible Trajectories for Quadrotors Using Safe Flight Corridors in 3-D Complex Environments Planning dynamically feasible trajectories for quadrotors using safe flight corridors in 3-D complex ...

    arXiv 2017

  32. [32]

    long_sensor-based APACrefauthors Long, K. , Yi, Y. , Dai, Z. , Herbert, S. , Cortés, J. Atanasov, N. APACrefauthors \ 2026 . Sensor-Based Distributionally Robust Control for Safe Robot Navigation in Dynamic Environments Sensor-based distributionally robust control for safe robot navigation in dynamic environments . The International Journal of Robotics Re...

    work page doi:10.1177/02783649251352000 2026

  33. [33]

    \ Tedrake, R

    majumdar2017funnel APACrefauthors Majumdar, A. \ Tedrake, R. APACrefauthors \ 2017 . Funnel libraries for real-time robust feedback motion planning Funnel libraries for real-time robust feedback motion planning . The International Journal of Robotics Research 36 8 947--982, APACrefDOI doi:10.1177/0278364917712421 APACrefDOI

    doi:10.1177/0278364917712421 2017

  34. [34]

    , Nobel, P

    marcucci2024boxes APACrefauthors Marcucci, T. , Nobel, P. , Tedrake, R. Boyd, S. APACrefauthors \ 2024 . Fast Path Planning Through Large Collections of Safe Boxes Fast path planning through large collections of safe boxes . IEEE Transactions on Robotics 40 3795-3811, APACrefDOI doi:10.1109/TRO.2024.3434168 APACrefDOI

    arXiv 2024

  35. [35]

    , Petersen, M

    Marcucci_motion_planning APACrefauthors Marcucci, T. , Petersen, M. , von Wrangel, D. Tedrake, R. APACrefauthors \ 2023 . Motion planning around obstacles with convex optimization Motion planning around obstacles with convex optimization . Science Robotics 8 84 eadf7843, APACrefDOI doi:10.1126/scirobotics.adf7843 APACrefDOI

    doi:10.1126/scirobotics.adf7843 2023

  36. [36]

    \ Kumar, V

    MinSnapTrajectory APACrefauthors Mellinger, D. \ Kumar, V. APACrefauthors \ 2011 . Minimum Snap Trajectory Generation and Control for Quadrotors Minimum snap trajectory generation and control for quadrotors . IEEE International Conference on Robotics and Automation (ICRA) IEEE International Conference on Robotics and Automation (ICRA) \ ( 2520-2525)

    2011

  37. [37]

    , Srinivasan, P.P

    nerf2020 APACrefauthors Mildenhall, B. , Srinivasan, P.P. , Tancik, M. , Barron, J.T. , Ramamoorthi, R. Ng, R. APACrefauthors \ 2020 . NeRF : Representing Scenes as Neural Radiance Fields for View Synthesis NeRF : Representing scenes as neural radiance fields for view synthesis . European Conference on Computer Vision (ECCV) European Conference on Compute...

    2020

  38. [38]

    , Oleynikova, H

    millane_nvblox_2024 APACrefauthors Millane, A. , Oleynikova, H. , Wirbel, E. , Steiner, R. , Ramasamy, V. , Tingdahl, D. Siegwart, R. APACrefauthors \ 2024 . nvblox : GPU -Accelerated Incremental Signed Distance Field Mapping nvblox : GPU -accelerated incremental signed distance field mapping . IEEE International Conference on Robotics and Automation (ICR...

    2024

  39. [39]

    , Watterson, M

    mohta_gps_denied APACrefauthors Mohta, K. , Watterson, M. , Mulgaonkar, Y. , Liu, S. , Qu, C. , Makineni, A. Kumar, V. APACrefauthors \ 2018 . Fast, autonomous flight in GPS -denied and cluttered environments Fast, autonomous flight in GPS -denied and cluttered environments . Journal of Field Robotics 35 1 101-120, APACrefDOI doi:https://doi.org/10.1002/r...

    work page doi:10.1002/rob.21774 2018

  40. [40]

    , Hehn, M

    mueller_motion_primitive APACrefauthors Mueller, M.W. , Hehn, M. D'Andrea, R. APACrefauthors \ 2015 . A Computationally Efficient Motion Primitive for Quadrocopter Trajectory Generation A computationally efficient motion primitive for quadrocopter trajectory generation . IEEE Transactions on Robotics 31 6 1294-1310, APACrefDOI doi:10.1109/TRO.2015.2479878...

    arXiv 2015

  41. [41]

    , Budsberg, J

    openvdb25 APACrefauthors Museth, K. , Budsberg, J. , Sirois-Vigneux, A. , Hurst, G. , Williams, F. , Pradhana, A. Avramoussis, N. APACrefauthors \ 2025 . OpenVDB OpenVDB . Proceedings of the Special Interest Group on Computer Graphics and Interactive Techniques Conference Courses . Proceedings of the Special Interest Group on Computer Graphics and Interac...

    2025

  42. [42]

    , Izadi, S

    newcombe_kinectfusion_2011 APACrefauthors Newcombe, R.A. , Izadi, S. , Hilliges, O. , Molyneaux, D. , Kim, D. , Davison, A.J. Fitzgibbon, A. APACrefauthors \ 2011 . KinectFusion : Real-Time Dense Surface Mapping and Tracking KinectFusion : Real-time dense surface mapping and tracking . IEEE International Symposium on Mixed and Augmented Reality (ISMAR) IE...

    2011

  43. [43]

    , Taylor, Z

    oleynikova_voxblox_2017 APACrefauthors Oleynikova, H. , Taylor, Z. , Fehr, M. , Siegwart, R. Nieto, J. APACrefauthors \ 2017 . Voxblox : Incremental 3D E uclidean Signed Distance Fields for On-Board MAV Planning Voxblox : Incremental 3D E uclidean signed distance fields for on-board MAV planning . IEEE/RSJ International Conference on Intelligent Robots an...

    2017

  44. [44]

    , Clegg, A

    ortiz_isdf_2022 APACrefauthors Ortiz, J. , Clegg, A. , Dong, J. , Sucar, E. , Novotny, D. , Zollhoefer, M. Mukadam, M. APACrefauthors \ 2022 . iSDF : Real-Time Neural Signed Distance Fields for Robot Perception iSDF : Real-time neural signed distance fields for robot perception . Robotics: Science and Systems (RSS) . Robotics: Science and Systems (RSS)

    2022

  45. [45]

    , Kompis, Y

    pan_voxfield_2022 APACrefauthors Pan, Y. , Kompis, Y. , Bartolomei, L. , Mascaro, R. , Stachniss, C. Chli, M. APACrefauthors \ 2022 . Voxfield : Non-Projective Signed Distance Fields for Online Planning and 3D Reconstruction Voxfield : Non-projective signed distance fields for online planning and 3D reconstruction . IEEE/RSJ International Conference on In...

    2022

  46. [46]

    , Zhong, X

    pan_pin-slam_2024 APACrefauthors Pan, Y. , Zhong, X. , Wiesmann, L. , Posewsky, T. , Behley, J. Stachniss, C. APACrefauthors \ 2024 . PIN-SLAM : LiDAR SLAM Using a Point-Based Implicit Neural Representation for Achieving Global Map Consistency PIN-SLAM : LiDAR SLAM using a point-based implicit neural representation for achieving global map consistency . I...

    arXiv 2024

  47. [47]

    , Florence, P

    park_deepsdf_2019 APACrefauthors Park, J.J. , Florence, P. , Straub, J. , Newcombe, R. Lovegrove, S. APACrefauthors \ 2019 . DeepSDF : Learning Continuous Signed Distance Functions for Shape Representation DeepSDF : Learning continuous signed distance functions for shape representation . IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR...

    2019

  48. [48]

    , Zhu, F

    ren2022bubbleplannerplanninghighspeed APACrefauthors Ren, Y. , Zhu, F. , Liu, W. , Wang, Z. , Lin, Y. , Gao, F. Zhang, F. APACrefauthors \ 2022 . Bubble Planner: Planning High-Speed Smooth Quadrotor Trajectories Using Receding Corridors Bubble planner: Planning high-speed smooth quadrotor trajectories using receding corridors . IEEE/RSJ International Conf...

    2022

  49. [49]

    , Zhu, F

    ren2025safety APACrefauthors Ren, Y. , Zhu, F. , Lu, G. , Cai, Y. , Yin, L. , Kong, F. Zhang, F. APACrefauthors \ 2025 . Safety-assured high-speed navigation for MAV s Safety-assured high-speed navigation for MAV s . Science Robotics 10 98 eado6187, APACrefDOI doi:10.1126/scirobotics.ado6187 APACrefDOI

    work page doi:10.1126/scirobotics.ado6187 2025

  50. [50]

    , Bry, A

    Richter2016 APACrefauthors Richter, C. , Bry, A. Roy, N. APACrefauthors \ 2016 . Polynomial Trajectory Planning for Aggressive Quadrotor Flight in Dense Indoor Environments Polynomial trajectory planning for aggressive quadrotor flight in dense indoor environments . M. Inaba\ P. Corke\ ( ), Robotics Research: The 16th International Symposium ISRR Robotics...

    doi:10.1007/9 2016

  51. [51]

    , Ware, J

    ryll_efficient_traj_planning APACrefauthors Ryll, M. , Ware, J. , Carter, J. Roy, N. APACrefauthors \ 2019 . Efficient Trajectory Planning for High Speed Flight in Unknown Environments Efficient trajectory planning for high speed flight in unknown environments . International Conference on Robotics and Automation (ICRA) International Conference on Robotic...

    2019

  52. [52]

    , Steinweg, M

    song_rl_drone_racing APACrefauthors Song, Y. , Steinweg, M. , Kaufmann, E. Scaramuzza, D. APACrefauthors \ 2021 . Autonomous Drone Racing with Deep Reinforcement Learning Autonomous drone racing with deep reinforcement learning . IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) IEEE/RSJ International Conference on Intelligent Rob...

    2021

  53. [53]

    , Whelan, T

    replica19arxiv APACrefauthors Straub, J. , Whelan, T. , Ma, L. , Chen, Y. , Wijmans, E. , Green, S. Newcombe, R. APACrefauthors \ 2019 . The R eplica Dataset: A Digital Replica of Indoor Spaces. The R eplica dataset: A digital replica of indoor spaces. Preprint at https://arxiv.org/abs/1906.05797

    Pith/arXiv arXiv 2019

  54. [54]

    , Moll, M

    ompl_2012 APACrefauthors Sucan, I.A. , Moll, M. Kavraki, L.E. APACrefauthors \ 2012 . The Open Motion Planning Library The Open Motion Planning Library . IEEE Robotics and Automation Magazine 19 4 72-82, APACrefDOI doi:10.1109/MRA.2012.2205651 APACrefDOI

    arXiv 2012

  55. [55]

    , Litalien, J

    takikawa_lod_2021 APACrefauthors Takikawa, T. , Litalien, J. , Yin, K. , Kreis, K. , Loop, C. , Nowrouzezahrai, D. Fidler, S. APACrefauthors \ 2021 . Neural Geometric Level of Detail: Real-Time Rendering with Implicit 3D Shapes Neural geometric level of detail: Real-time rendering with implicit 3D shapes . IEEE/CVF Conference on Computer Vision and Patter...

    2021

  56. [56]

    , Cao, H

    miso2025 APACrefauthors Tian, Y. , Cao, H. , Kim, S. Atanasov, N. APACrefauthors \ 2025 . MISO : Multiresolution Submap Optimization for Efficient Globally Consistent Neural Implicit Reconstruction MISO : Multiresolution submap optimization for efficient globally consistent neural implicit reconstruction . Robotics: Science and Systems (RSS) . Robotics: S...

    2025

  57. [57]

    \ How, J.P

    tordesillas2022faster APACrefauthors Tordesillas, J. \ How, J.P. APACrefauthors \ 2022 . FASTER : Fast and Safe Trajectory Planner for Navigation in Unknown Environments FASTER : Fast and safe trajectory planner for navigation in unknown environments . IEEE Transactions on Robotics 38 2 922-938, APACrefDOI doi:10.1109/TRO.2021.3100142 APACrefDOI

    arXiv 2022

  58. [58]

    APACrefauthors \ 1967

    varadhan_behavior_1967 APACrefauthors Varadhan, S.R.S. APACrefauthors \ 1967 . On the behavior of the fundamental solution of the heat equation with variable coefficients On the behavior of the fundamental solution of the heat equation with variable coefficients . Communications on Pure and Applied Mathematics 20 2 431--455, APACrefDOI doi:10.1002/cpa.316...

    doi:10.1002/cpa.3160200210 1967

  59. [59]

    , Garg, S

    hio-sdf_2024 APACrefauthors Vasilopoulos, V. , Garg, S. , Huh, J. , Lee, B. Isler, V. APACrefauthors \ 2024 . HIO-SDF : Hierarchical Incremental Online Signed Distance Fields HIO-SDF : Hierarchical incremental online signed distance fields . IEEE International Conference on Robotics and Automation (ICRA) IEEE International Conference on Robotics and Autom...

    2024

  60. [60]

    , Liu, L

    wang_neus_2021 APACrefauthors Wang, P. , Liu, L. , Liu, Y. , Theobalt, C. , Komura, T. Wang, W. APACrefauthors \ 2021 . NeuS : Learning Neural Implicit Surfaces by Volume Rendering for Multi-View Reconstruction NeuS : Learning neural implicit surfaces by volume rendering for multi-view reconstruction . International Conference on Neural Information Proces...

    2021

  61. [61]

    , Wang, C

    wang_hotspot_2024 APACrefauthors Wang, Z. , Wang, C. , Yoshino, T. , Tao, S. , Fu, Z. Li, T M. APACrefauthors \ 2025 . HotSpot : Signed Distance Function Optimization with an Asymptotically Sufficient Condition HotSpot : Signed distance function optimization with an asymptotically sufficient condition . IEEE/CVF Conference on Computer Vision and Pattern R...

    2025

  62. [62]

    , Zhou, X

    MINCO APACrefauthors Wang, Z. , Zhou, X. , Xu, C. Gao, F. APACrefauthors \ 2022 . Geometrically Constrained Trajectory Optimization for Multicopters Geometrically constrained trajectory optimization for multicopters . IEEE Transactions on Robotics 38 5 3259-3278, APACrefDOI doi:10.1109/TRO.2022.3160022 APACrefDOI

    arXiv 2022

  63. [63]

    , Lee, K.M.B

    lan2021loggpis APACrefauthors Wu, L. , Lee, K.M.B. , Liu, L. Vidal-Calleja, T. APACrefauthors \ 2021 . Faithful E uclidean Distance Field from Log- G aussian Process Implicit Surfaces Faithful E uclidean distance field from log- G aussian process implicit surfaces . IEEE Robotics and Automation Letters 6 2 2461-2468, APACrefDOI doi:10.1109/LRA.2021.306135...

    arXiv 2021

  64. [64]

    , Le Gentil, C

    wu_vdb-gpdf_2025 APACrefauthors Wu, L. , Le Gentil, C. Vidal-Calleja, T. APACrefauthors \ 2025 . VDB-GPDF : Online G aussian Process Distance Field with VDB Structure VDB-GPDF : Online G aussian process distance field with VDB structure . IEEE Robotics and Automation Letters 10 1 374-381, APACrefDOI doi:10.1109/LRA.2024.3505814 APACrefDOI

    arXiv 2025

  65. [65]

    , Camurri, M

    newercollege2021 APACrefauthors Zhang, L. , Camurri, M. , Wisth, D. Fallon, M. APACrefauthors \ 2021 . Multi-Camera LiDAR Inertial Extension to the N ewer C ollege Dataset. Multi-camera LiDAR inertial extension to the N ewer C ollege dataset. Preprint at https://arxiv.org/abs/2112.08854

    Pith/arXiv arXiv 2021

  66. [66]

    , Gao, F

    zhou_robust_efficient APACrefauthors Zhou, B. , Gao, F. , Wang, L. , Liu, C. Shen, S. APACrefauthors \ 2019 . Robust and Efficient Quadrotor Trajectory Generation for Fast Autonomous Flight Robust and efficient quadrotor trajectory generation for fast autonomous flight . IEEE Robotics and Automation Letters 4 4 3529-3536, APACrefDOI doi:10.1109/LRA.2019.2...

    arXiv 2019