Pith. sign in

REVIEW 4 major objections 5 minor 70 references

Lifelong Localization in Dynamic Indoor Environments Combining Odometry with Sparse Distance Sampling

T0 review · 4 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read A robot can localize itself with just 16 distance samples and odometry, with provable convergence to its true pose even in dynamic indoor environments.

desk verdict A genuinely useful engineering extension of the authors' own sparse-distance localization, but the headline 'provably converges' claim is not actually proven, and the dynamic guarantee rests on a heuristic fitted in one building. read the letter →

arxiv 2607.17852 v1 pith:LNO6UAXN submitted 2026-07-20 cs.RO

Michael M. Bilevich , Tomer Buber , Dan Halperin This is my paper · ORCID
classification cs.RO MSC 68T4068U05
keywords robotlocalizationsparsedistancesamplingodometryfusionkidnappedproblemlifelongsubdivisionsearchdynamicenvironmentsBayesianfiltering
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

This paper claims that a planar robot navigating a known indoor map can determine its location reliably using only a handful of distance measurements—16 in the experiments—together with odometry, instead of a full LiDAR scan. The method solves the kidnapped robot problem from scratch at every iteration, meaning it can recover from sudden displacement without any explicit detection routine. The authors prove that the estimate converges to the robot's true pose when the environment is static, and they argue the same guarantee holds in dynamic environments as long as the nature of the changes has been learned correctly. This matters because sparse sampling promises cheaper sensors, better privacy, lower storage, and reduced transmission bandwidth, making lifelong localization feasible for low-cost robots.

What carries the argument

The central mechanism is the subdivision search of the preimage of a sparse distance measurement: a quadtree-like recursive partition of the configuration space (position plus orientation) in which a voxel is kept only if at least k' of the k distance rays could intersect the map boundary within an error bound ε. The key heuristic is the estimated k' value, f̃_k'(dist), a fitted decision stump based on the robot's distance to the map boundary, which determines how many measurements must agree with the static map. This candidate set is then fused with dead-reckoning odometry through a Bayesian belief update, where each candidate pose's likelihood is propagated using a Gaussian transition mode

What would settle it

Run the algorithm in an environment whose room geometry and obstacle behavior differ from the training office, with many dynamic obstacles positioned close to walls so that the true k'/k ratio falls below the heuristic's threshold, and check whether the reported pose stays within δ of ground truth; a divergence would show that the k' heuristic is not a universally correct lower bound.

Watch

Extended reading notes

Core claim

The paper's central claim is that localizing a robot in a known planar indoor map reduces to intersecting the fibers of a distance-measurement function, and that this intersection can be approximated efficiently by a subdivision search over the configuration space. The method, called SDSL, requires only k distance samples and odometry; it provably outputs at least one pose within δ of the ground truth provided that the heuristic estimating k'—the number of rays that hit static map features—is a correct lower bound for every voxel. The authors further show that fusing this candidate pose set with odometry through a Bayesian filter yields lifelong localization that is inherently robust to kidn

Load-bearing premise

The convergence and robustness guarantee rests on the learned estimate of k'—the number of rays that hit static map features—being a correct lower bound for every voxel; if this heuristic is wrong in a new environment, the algorithm can discard the true-pose voxel.

Editorial extensions

If this is right

  • A robot can achieve SLAM-comparable localization accuracy using only a sparse set of distance samples, greatly reducing sensor cost and data bandwidth.
  • The kidnapped robot problem is solved implicitly at every iteration, eliminating the need for separate kidnapping detection or re-sampling routines.
  • In static environments, the algorithm is guaranteed to always report a pose close to the ground truth, irrespective of how wrong the initial guess is.
  • In dynamic environments, the same guarantee holds provided the statistical nature of the changes is learned correctly, making the method suitable for lifelong operation.
  • The output-sensitive complexity means the method's runtime scales with the number of geometrically feasible poses, which can be far smaller than the configuration space volume.

Reading between the lines

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

  • The distance-to-boundary heuristic is environment-specific: the paper's own kitchenette data yield a different decision stump, suggesting that transferring the method to a new building requires re-learning f̃_k' or the guarantee may fail.
  • The method could likely be adapted to other sparse range sensors (e.g., ultrasonic or single-beam lidar) since it only needs a few distance measurements, expanding its applicability to even cheaper hardware.
  • In highly symmetric environments, the pose candidate set may contain multiple distinct hypotheses, and the guarantee only ensures at least one is correct; active motion or additional constraints would be needed to disambiguate them.
  • The convergence guarantee in dynamic environments is conditional on the learned model being a correct lower bound, so online adaptation of k' during operation would be a natural extension to handle unforeseen dynamics.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes SDSL, a lifelong localization method for planar robots that combines odometry with k=16 sparse distance samples. Given a pre-existing map, the method computes candidate poses by a subdivision search over the configuration space, intersecting fibers of the distance measurement function (Section 4, Algorithm 1). A heuristic f̃_k'(dist) (Eq. 1) estimates the number of rays expected to hit static map features, and voxels that do not match at least k' samples are pruned. The candidate set is then fused with odometry via a Bayesian update (Section 5, Algorithm 3), and the final pose is selected by clustering and taking a weighted cluster center (Section 3.3). The paper claims that this procedure provably converges to the robot's ground-truth pose in static environments, and also in dynamic environments when the dynamic nature has been correctly learned (Abstract, Section 6). Experiments on a physical iRobot Create 3 in three real indoor environments (lab446, fl4, apt) compare SDSL against nav2-amcl, reporting comparable or better localization accuracy and faster recovery in kidnapped-robot scenarios (Section 7).

Significance. If the convergence claim were substantiated, this would be a valuable contribution: localization with only sixteen distance samples could reduce sensor cost and bandwidth, and the deterministic candidate-generation approach offers a principled alternative to particle filtering. The paper's strengths include a real-robot evaluation in multiple environments, a comparison with a standard AMCL baseline, and an open-source C++ library with Python bindings and ROS2 packages (Section 7.1). The complexity comparison in Corollary 6.1 is also a useful theoretical observation. However, the central theoretical claim is currently not supported by the theorems as stated. The formal guarantee (Theorem 6.3) covers only the candidate set produced by Algorithm 1, is conditional on a k' estimator whose correctness is not established, and says nothing about the fused belief or the final reported pose. The paper's main contribution is therefore better described as a complete candidate-generation method with a plausible fusion heuristic, rather than a provably convergent localization algorithm.

major comments (4)
  1. [§6, Theorem 6.3 vs. Abstract and §3.3] The abstract claims the method 'provably converges to the robot's ground truth pose,' but Theorem 6.3 only guarantees that Algorithm 1 (the SDSL candidate-set generator) outputs at least one pose δ-close to the ground truth, and only under the assumption that EstimateKPrime is a correct lower bound. No theorem analyzes Algorithm 3's Bayesian belief update or the cluster-selection step at the end of §3.3. In an environment with symmetries or perceptual aliasing, multiple candidates can remain consistent with all distance samples and odometry indefinitely; without an observability or identifiability argument, the reported pose could converge to the wrong mode. The empirical results in Table 1 are good but do not fill this proof gap. Please either provide a convergence analysis for the fused estimate under explicit assumptions, or revise the claims to state precisely what is proven.
  2. [§4.1, Eq. (1) and §7.4] Theorem 6.3 is conditional on EstimateKPrime returning a correct lower bound on the number of rays hitting static map features. The only supplied estimator is the fitted decision stump f̃_k'(dist) from Eq. (1), with threshold 0.975 m and ratios 0.8/0.7. Section 7.4 shows that the same procedure fitted to the kitchenette of fl4 yields a different decision stump (threshold 0.75 m, ratios 0.8086/0.8279). The paper does not establish that f̃_k' is a lower bound in unseen environments, nor that the voxel evaluation (taking the minimum over vertices and center) is a lower bound for every pose inside the voxel. A voxel can straddle the threshold, so the minimum of a step function at a few points is not necessarily conservative for the whole voxel. This is a load-bearing gap: if the estimate is wrong in a new environment, Algorithm 1 can discard the true-pose voxel and the guarantee fails. Pleas
  3. [Algorithm 2] The pseudocode's final line is 'return B∩W ≠ ∅', but the geometric correctness argument in §4 and the preceding text require intersection with the boundary: 'if the intersection F_{d,g}(V)∩∂W ≠ ∅'. If Algorithm 2 is implemented literally, the predicate becomes true for any voxel whose bounding box overlaps the map interior, regardless of whether a measured distance d can be realized by a ray hitting ∂W. This would destroy the pruning in Algorithm 1 and invalidate Theorem 6.2's convergence to the fiber intersection. Please correct the pseudocode to test B∩∂W (or otherwise clarify the intended test) and confirm that the implementation matches the corrected version.
  4. [§5, Algorithm 3] The Bayesian fusion step is stated but never analyzed. Equation (3) and Algorithm 3 define a belief update, but there is no proof that Bel concentrates on the true candidate, that the normalized weights converge, or that the cluster center reported at the end of §3.3 is the δ-close pose guaranteed by Theorem 6.3. The transition model uses a Gaussian with standard deviation ε, but ε is also used as the measurement error bound in Algorithm 1; the relationship between these two roles is not discussed. Please add an analysis of the fusion step showing convergence to the true mode under stated conditions, or explicitly delineate the candidates-generation guarantee from the empirical fusion behavior.
minor comments (5)
  1. [Theorem 6.3] The theorem statement says 'correct lower bound to the actual number of perceived dynamic obstacles' — this should presumably read 'static map features', since k' is defined in Section 4 as the number of samples corresponding to the pre-determined map. Please correct the terminology.
  2. [Section 7.4] There is a typo: 'interdependently evaluated' should likely be 'independently evaluated'.
  3. [Algorithm 1 / Section 5] Algorithm 1 returns a set Q of voxels, while Section 5 treats X_n as a set of configurations. The conversion from voxels to pose candidates via voxel centers is mentioned only in passing (§4). Please make this conversion explicit so that the input to Algorithm 3 is unambiguous.
  4. [Abstract and §7.3] The claim of solving the kidnapped robot problem 'in real time' should be qualified: Section 7.3 reports convergence times up to 15.7 seconds on fl4. This may be acceptable, but the phrase 'real time' is not defined.
  5. [Table 1] The header formatting 'Oursamcl' and the lack of standard deviation or per-lap breakdown make the comparison harder to interpret. Please include more statistical detail, e.g., mean ± std over the five laps.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the fitted k' heuristic is an explicit empirical input, the formal guarantee is conditional on it, and the core SDSL algorithm is inherited from prior work rather than being re-derived from the target claim.

full rationale

The paper's central derivation chain is not circular. The static convergence claim inherits from the authors' prior subdivision-search work [42], but that is prior peer-reviewed work used as a building block, not an unverified self-citation invoked to forbid alternatives. The dynamic guarantee in Theorem 6.3 is explicitly conditional: "Assuming that the k' estimate value of EstimateKPrime(V) is a correct lower bound to the actual number of perceived dynamic obstacles for each voxel V, Algorithm 1 is guaranteed to output at least one pose q which is δ-close to the ground truth location." The implementation of EstimateKPrime is the empirically fitted decision stump f̃_k'(dist) from Eq. (1), fitted on lab446 data and checked against kitchenette data in Section 7.4. This is a fitted input used in an algorithm, not a fitted parameter renamed as a prediction of the localization result; the localization accuracy is measured against landmarks and compared with nav2-amcl. The kitchenette decision stump differs (threshold 0.75 m vs 0.975 m), and the paper honestly notes in Section 8 that "when the robot faces many dynamic obstacles in extreme cases, we may temporarily lose its location." That is a transfer/robustness limitation, not a circular derivation. The main proof gap is that the abstract's "provably converges to the robot's ground truth pose" is not fully established for the fused output: Theorem 6.3 only guarantees that Algorithm 1 outputs a δ-close candidate, while the Bayesian fusion in Algorithm 3 and the final cluster selection are not analyzed with a convergence theorem. This is a correctness and rigor concern, but it is not an instance of a prediction reducing to its inputs by construction. No equation in the paper is shown to be equivalent to its own input, and no fitted value is presented as a derivation of the target result. Therefore the appropriate finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central static guarantee relies on standard geometry and the k-k' gap model; the dynamic guarantee additionally relies on a fitted heuristic. No new physical entities are introduced.

free parameters (4)
  • f̃_k′ heuristic threshold and ratios = dist < 0.975 m → 0.8; else → 0.7
    Fitted via a decision stump to recorded LiDAR data in lab446 (Section 7.4), then rounded down and used globally. The kitchenette fit gives a different threshold (0.75 m) and different ratios, so the reported values are environment-specific.
  • δ (voxel approximation precision) = not reported
    Required by Algorithm 1 to terminate; only constrained to be smaller than the desired accuracy. The experimental value is not given, which affects candidate count and runtime.
  • ε (measurement error bound / Gaussian std dev) = not reported
    Used in VoxelXPred to expand bounding boxes and in Algorithm 3 as the transition standard deviation; no value is specified in the experiments.
  • k (number of distance samples) = 16
    Chosen for the robot experiments ('merely a few (sixteen) distance samples'); the k' study samples k=4..20 but the final runs use 16. This is a design choice, not derived.
assumptions (5)
  • domain assumption The environment is a closed subset W⊂R² and the map is a good approximation of it, possibly with topological errors and dynamic obstacles.
    Sections 2.2 and 3.1 define the problem and justify matching distance samples against the map boundary ∂W.
  • domain assumption At least k' of the k range measurements correspond to static features in the precomputed map (the k-k' dynamic gap).
    Section 4 models dynamic obstacles this way; Theorem 6.3 assumes EstimateKPrime gives a correct lower bound on k'.
  • ad hoc to paper The specific heuristic f̃_k′(dist) in Eq. (1) is a correct lower bound for the environments tested.
    Learned from one office environment's recordings (Section 7.4) and asserted to be a lower bound; the paper's own kitchenette data produce a different decision stump.
  • domain assumption Odometry provides an estimate Un whose error is modeled as a zero-mean Gaussian with standard deviation ε.
    Section 5, Eq. (2) and Algorithm 3 use this standard model; it is not verified for the iRobot Create 3 used in experiments.
  • standard math Standard computational-geometry facts about subdivision search and Hausdorff dimension; correctness/complexity proofs are delegated to [42].
    Section 6 relies on these for Theorems 6.1–6.2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Lifelong Localization in Dynamic Indoor Environments Combining Odometry with Sparse Distance Sampling." pith.science (2026). https://pith.science/paper/LNO6UAXN

@misc{pith2026260717852,
  author       = {Pith},
  title        = {Pith review of: Lifelong Localization in Dynamic Indoor Environments Combining Odometry with Sparse Distance Sampling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LNO6UAXN}},
  note         = {Machine review of arXiv:2607.17852}
}
read the original abstract

Localization is a key task in robot navigation, and many techniques exist for it. In many plausible scenarios, a robot might face unforeseen, dynamic obstacles, rendering any pre-determined map inaccurate for localization. In this work, we propose a robust lifelong localization framework in dynamic planar indoor environments, using the robot's odometry and sparse distance sampling. We demonstrate how distance samples can be used to provide a robust prior on the robot's location. This technique can solve the kidnapped robot problem in real time, up to symmetries. Based on insights from real-world recorded data, we also account for dynamic obstacles. We then fuse this prior, over time, with the odometry to converge to the robot's location. A central property of our method is that it provably converges to the robot's ground truth pose even in large indoor environments when the environment is static. We further show that this guarantee also holds in dynamic environments, as long as the nature of those changes has been correctly learned. We demonstrate the effectiveness of our approach in different real-world indoor environments. In particular, we achieve a localization comparable to SLAM with merely a few (sixteen) distance samples, as opposed to the full LiDAR range. Sufficing with only sparse distance sampling is advantageous in terms of sensor cost, privacy, storage space, and transmission bandwidth.

Figures

Figures reproduced from arXiv: 2607.17852 by the authors.

Figure 1
Figure 1. A demonstration of our algorithm in different real-world scenarios. All [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. An overview of our method. On the left-hand side are the inputs to our [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Illustration of two poses with k = 6 distance measurements. The top point (whose rays are cyan) is closer to the environment’s wall than the bottom (whose rays are orange). Notice how the bottom point is more likely to see dynamic obstacles, as there is more space between the robot and the environment. For the top point, about half of its rays measure the wall, which is relatively close. that pose, the sensor whose … view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: An image of our iRobot Create 3 in a corridor of map [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: Visualization of the three chosen landmarks in each environment, circled [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

70 extracted references · 5 linked inside Pith

  1. [1]

    A Survey on Indoor Vehicle Localization Through RFID Technology,

    A. Motroni, A. Buffi, and P. Nepa, “A Survey on Indoor Vehicle Localization Through RFID Technology,”IEEE Access, vol. 9, pp. 17 921–17 942, 2021

    2021

  2. [2]

    Adaptive Kalman Fil- tering for GPS-based Mobile Robot Localization,

    G. Reina, A. Vargas, K. Nagatani, and K. Yoshida, “Adaptive Kalman Fil- tering for GPS-based Mobile Robot Localization,” in2007 IEEE International Workshop on Safety, Security and Rescue Robotics, Sep. 2007, pp. 1–6, iSSN: 2374-3247

    2007

  3. [3]

    Comparison of Indoor Robot Local- ization Techniques in the Absence of GPS,

    R. Vincent, B. Limketkai, and M. Eriksen, “Comparison of Indoor Robot Local- ization Techniques in the Absence of GPS,” inDetection and Sensing of Mines, Explosive Objects, and Obscured Targets XV, vol. 7664. SPIE, Apr. 2010, pp. 606–610

    2010

  4. [4]

    Sensor Fusion of INS, Odometer and GPS for Robot Localization,

    S. Yousuf and M. B. Kadri, “Sensor Fusion of INS, Odometer and GPS for Robot Localization,” in2016 IEEE Conference on Systems, Process and Control (ICSPC), Dec. 2016, pp. 118–123

    2016

  5. [5]

    A Survey of Robot Swarms’ Relative Localization Method,

    S. Chen, D. Yin, and Y. Niu, “A Survey of Robot Swarms’ Relative Localization Method,”Sensors, vol. 22, no. 12, p. 4424, Jan. 2022

    2022

  6. [6]

    A Survey of Recent Indoor Localization Scenarios and Methodologies,

    T. Yang, A. Cabani, and H. Chafouk, “A Survey of Recent Indoor Localization Scenarios and Methodologies,”Sensors, vol. 21, no. 23, p. 8086, Jan. 2021

    2021

  7. [7]

    Application of QR Code for Localization and Navigation of Indoor Mobile Robot,

    S.-H. Bach, P.-B. Khoi, and S.-Y. Yi, “Application of QR Code for Localization and Navigation of Indoor Mobile Robot,”IEEE Access, vol. 11, pp. 28 384– 28 390, 2023

    2023

  8. [8]

    A Comprehensive Survey of Indoor Localization Methods Based on Computer Vision,

    A. Morar, A. Moldoveanu, I. Mocanu, F. Moldoveanu, I. E. Radoi, V. Asavei, A. Gradinaru, and A. Butean, “A Comprehensive Survey of Indoor Localization Methods Based on Computer Vision,”Sensors, vol. 20, no. 9, p. 2641, Jan. 2020

    2020

Show all 70 references
  1. [9]

    Indoor Localization of Mobile Robots Through QR Code Detection and Dead Reckoning Data Fusion,

    P. Nazemzadeh, D. Fontanelli, D. Macii, and L. Palopoli, “Indoor Localization of Mobile Robots Through QR Code Detection and Dead Reckoning Data Fusion,” IEEE/ASME Transactions on Mechatronics, vol. 22, no. 6, pp. 2588–2599, Dec. 2017

    2017

  2. [10]

    Performance Bounds in Positioning with the VIVE Lighthouse System,

    M. Greiff, A. Robertsson, and K. Berntorp, “Performance Bounds in Positioning with the VIVE Lighthouse System,” in2019 22th International Conference on Information Fusion (FUSION), Jul. 2019, pp. 1–8

    2019

  3. [11]

    DronOS: a flexible open- source prototyping framework for interactive drone routines,

    M. Hoppe, M. Burger, A. Schmidt, and T. Kosch, “DronOS: a flexible open- source prototyping framework for interactive drone routines,” inProceedings of the 18th International Conference on Mobile and Ubiquitous Multimedia, ser. MUM ’19. New York, NY, USA: Association for Compu...

    2019

  4. [12]

    Lighthouse Positioning System: Dataset, Accuracy, and Precision for UA V Research,

    A. Taffanel, B. Rousselot, J. Danielsson, K. McGuire, K. Richardsson, M. Elias- son, T. Antonsson, and W. H¨ onig, “Lighthouse Positioning System: Dataset, Accuracy, and Precision for UA V Research,” Apr. 2021, arXiv:2104.11523 [cs]

    2021 arXiv

  5. [13]

    Global Self-Localization of a Robot in Underground Mines,

    P. Debanne, J.-V. Herve, and P. Cohen, “Global Self-Localization of a Robot in Underground Mines,” inComputational Cybernetics and Simulation 1997 IEEE International Conference on Systems, Man, and Cybernetics, vol. 5, Oct. 1997, pp. 4400–4405 vol.5, iSSN: 1062-922X

    1997

  6. [14]

    Robust Localization for Underground Mining Vehicles: An Application in a Room and Pillar Mine,

    F. Inostroza, I. Parra-Tsunekawa, and J. Ruiz-del Solar, “Robust Localization for Underground Mining Vehicles: An Application in a Room and Pillar Mine,” Sensors, vol. 23, no. 19, p. 8059, Jan. 2023

    2023

  7. [15]

    Three-dimensional Lo- calization and Mapping for Mobile Robot in Disaster Environments,

    K. Nagatani, H. Ishida, S. Yamanaka, and Y. Tanaka, “Three-dimensional Lo- calization and Mapping for Mobile Robot in Disaster Environments,” inProceed- ings 2003 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2003) (Cat. No.03CH37453), vol. 4, Oct. ...

    2003

  8. [16]

    Localization Using Triangulation in Swarms of Autonomous Rescue Robots,

    D. P. Stormont and A. Kutiyanawala, “Localization Using Triangulation in Swarms of Autonomous Rescue Robots,” in2007 IEEE International Workshop on Safety, Security and Rescue Robotics, Sep. 2007, pp. 1–6, iSSN: 2374-3247

    2007

  9. [17]

    A Survey of State-of-the-art on Visual SLAM,

    I. Abaspur Kazerouni, L. Fitzgerald, G. Dooly, and D. Toal, “A Survey of State-of-the-art on Visual SLAM,”Expert Systems with Applications, vol. 205, p. 117734, Nov. 2022

    2022

  10. [18]

    The SLAM problem: a survey,

    J. Aulinas, Y. Petillot, J. Salvi, and X. Llado, “The SLAM problem: a survey,” inArtificial Intelligence Research and Development. IOS Press, 2008, pp. 363– 371

    2008

  11. [19]

    A Comparative Survey of LiDAR-SLAM and LiDAR based Sensor Tech- nologies,

    M. U. Khan, S. A. A. Zaidi, A. Ishtiaq, S. U. R. Bukhari, S. Samer, and A. Far- man, “A Comparative Survey of LiDAR-SLAM and LiDAR based Sensor Tech- nologies,” in2021 Mohammad Ali Jinnah University International Conference on Computing (MAJICC), Jul. 2021, pp. 1–8

    2021

  12. [20]

    3D LiDAR SLAM: A Survey,

    Y. Zhang, P. Shi, and J. Li, “3D LiDAR SLAM: A Survey,”The Photogrammet- ric Record, vol. 39, no. 186, pp. 457–517, 2024

    2024

  13. [21]

    Indoor Drone Positioning: Accuracy and Cost Trade-Off for Sensor Fusion,

    J. V.-V. Gerwen, K. Geebelen, J. Wan, W. Joseph, J. Hoebeke, and E. De Poorter, “Indoor Drone Positioning: Accuracy and Cost Trade-Off for Sensor Fusion,”IEEE Transactions on Vehicular Technology, vol. 71, no. 1, pp. 961–974, Jan. 2022

    2022

  14. [22]

    Localization of Inexpensive Robots with Low-Bandwidth Sensors,

    S. Wang, F. Colas, M. Liu, F. Mondada, and S. Magnenat, “Localization of Inexpensive Robots with Low-Bandwidth Sensors,” inDistributed Autonomous Robotic Systems: The 13th International Symposium, R. Groß, A. Kolling, S. Berman, E. Frazzoli, A. Martinoli, F. Matsuno, and M. Ga...

    2018

  15. [23]

    Detection of Privacy- Sensitive Situations for Social Robots in Smart Homes,

    F. E. Fernandes, G. Yang, H. M. Do, and W. Sheng, “Detection of Privacy- Sensitive Situations for Social Robots in Smart Homes,” in2016 IEEE Interna- tional Conference on Automation Science and Engineering (CASE), Aug. 2016, pp. 727–732, iSSN: 2161-8089

    2016

  16. [24]

    The Robot Privacy Paradox: Understanding How Privacy Concerns Shape Intentions to Use Social Robots,

    C. Lutz and A. Tamo-Larrieux, “The Robot Privacy Paradox: Understanding How Privacy Concerns Shape Intentions to Use Social Robots,”Human-Machine Communication, vol. 1, pp. 87–111, Jan. 2020

    2020

  17. [25]

    State Estimation Algorithms for Localization: A Survey,

    J. M. Pak and C. K. Ahn, “State Estimation Algorithms for Localization: A Survey,”International Journal of Control, Automation and Systems, vol. 21, no. 9, pp. 2771–2781, Sep. 2023

    2023

  18. [26]

    Dead reckoning for a lunar rover on uneven terrain,

    Y. Fuke and E. Krotkov, “Dead reckoning for a lunar rover on uneven terrain,” inProceedings of IEEE International Conference on Robotics and Automation, vol. 1, Apr. 1996, pp. 411–416 vol.1, iSSN: 1050-4729

    1996

  19. [27]

    Analysis of position estimation in a dead reckoning navigation robot,

    K. Halvorsen, “Analysis of position estimation in a dead reckoning navigation robot,” Master’s thesis, NTNU, 2021

    2021

  20. [28]

    DeRO: Dead Reckoning Based on Radar Odometry With Accelerometers Aided for Robot Localization,

    H. V. Do, Y. Hun Kim, J. H. Lee, M. Ho Lee, and J. W. Song, “DeRO: Dead Reckoning Based on Radar Odometry With Accelerometers Aided for Robot Localization,” in2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Oct. 2024, pp. 8547–8554, iSSN: 2153-0866

    2024

  21. [29]

    Model-Based Deep Learning for Low-Cost IMU Dead Reckoning of Wheeled Mobile Robot,

    F. Guo, H. Yang, X. Wu, H. Dong, Q. Wu, and Z. Li, “Model-Based Deep Learning for Low-Cost IMU Dead Reckoning of Wheeled Mobile Robot,”IEEE Transactions on Industrial Electronics, vol. 71, no. 7, pp. 7531–7541, Jul. 2024

    2024

  22. [30]

    Lifelong Localization in Changing Environments,

    G. D. Tipaldi, D. Meyer-Delius, and W. Burgard, “Lifelong Localization in Changing Environments,”The International Journal of Robotics Research, vol. 32, no. 14, pp. 1662–1678, Dec. 2013

    2013

  23. [31]

    Summary Maps for Lifelong Visual Localization,

    P. M¨ uhlfellner, M. B¨ urki, M. Bosse, W. Derendarz, R. Philippsen, and P. Furgale, “Summary Maps for Lifelong Visual Localization,”Journal of Field Robotics, vol. 33, no. 5, pp. 561–590, 2016

    2016

  24. [32]

    A General Framework for Lifelong Localization and Mapping in Changing Environment,

    M. Zhao, X. Guo, L. Song, B. Qin, X. Shi, G. H. Lee, and G. Sun, “A General Framework for Lifelong Localization and Mapping in Changing Environment,” in2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Sep. 2021, pp. 3305–3312

    2021

  25. [33]

    A Systematic Literature Review on Long-term Localization and Mapping for Mobile Robots,

    R. B. Sousa, H. M. Sobreira, and A. P. Moreira, “A Systematic Literature Review on Long-term Localization and Mapping for Mobile Robots,”Journal of Field Robotics, vol. 40, no. 5, pp. 1245–1322, 2023

    2023

  26. [34]

    LL-Localizer: A Life-Long Localization System based on Dynamic i-Octree,

    X. Li, S. Yuan, H. Cai, S. Lu, W. Wang, and J. Liu, “LL-Localizer: A Life-Long Localization System based on Dynamic i-Octree,” Apr. 2025, arXiv:2504.01583 [cs]

    2025 arXiv

  27. [35]

    LiLoc: Lifelong Localization Using Adaptive Submap Joining and Egocentric Factor Graph,

    Y. Fang, Y. Li, K. Qian, F. Tombari, Y. Wang, and G. H. Lee, “LiLoc: Lifelong Localization Using Adaptive Submap Joining and Egocentric Factor Graph,” in 2025 IEEE International Conference on Robotics and Automation (ICRA), May 2025, pp. 8041–8047

    2025

  28. [36]

    SLAM- RAMU: 3D LiDAR-IMU Lifelong SLAM with Relocalization and Autonomous Map Updating for Accurate and Reliable Navigation,

    B. Chen, X. Zhong, H. Xie, P. Peng, H. Hu, X. Zhong, and Q. Liu, “SLAM- RAMU: 3D LiDAR-IMU Lifelong SLAM with Relocalization and Autonomous Map Updating for Accurate and Reliable Navigation,”Industrial Robot: the international journal of robotics research and application, vol....

    2024

  29. [37]

    Guaranteed SLAM—An interval approach,

    M. Mustafa, A. Stancu, N. Delanoue, and E. Codres, “Guaranteed SLAM—An interval approach,”Robotics and Autonomous Systems, vol. 100, pp. 160–170, Feb. 2018

    2018

  30. [38]

    Guaranteed inter- val analysis localization for mobile robots,

    R. Guyonneau, S. Lagrange, L. Hardouin, and P. Lucidarme, “Guaranteed inter- val analysis localization for mobile robots,”Advanced Robotics, vol. 28, no. 16, pp. 1067–1077, Aug. 2014

    2014

  31. [39]

    Guaranteed 2D Pose Graph SLAM With Bounded Noises: An Efficient Interval Approach,

    Y. Song, H. Yang, L. Zhao, and S. Huang, “Guaranteed 2D Pose Graph SLAM With Bounded Noises: An Efficient Interval Approach,”IEEE Transactions on Automation Science and Engineering, pp. 1–12, 2025

    2025

  32. [40]

    Sensor Localization by Few Distance Measurements via the Intersection of Implicit Manifolds,

    M. M. Bilevich, S. M. LaValle, and D. Halperin, “Sensor Localization by Few Distance Measurements via the Intersection of Implicit Manifolds,” in2023 IEEE International Conference on Robotics and Automation (ICRA), May 2023, pp. 1912–1918

    2023

  33. [41]

    Localization in Dynamic Planar Environments Using Few Distance Measurements,

    M. M. Bilevich, S. Guini, and D. Halperin, “Localization in Dynamic Planar Environments Using Few Distance Measurements,” Aug. 2024, arXiv:2407.03219 [cs]

    2024 arXiv

  34. [42]

    Indoor Localization of UA Vs Using Only Few Measurements by Output-Sensitive Preimage Intersection,

    M. M. Bilevich, T. Buber, and D. Halperin, “Indoor Localization of UA Vs Using Only Few Measurements by Output-Sensitive Preimage Intersection,” in2025 IEEE International Conference on Robotics and Automation (ICRA), May 2025, pp. 5773–5780

    2025

  35. [43]

    Bayesian filtering: From Kalman filters to particle filters, and beyond,

    Z. Chen, “Bayesian filtering: From Kalman filters to particle filters, and beyond,” Statistics, vol. 182, no. 1, pp. 1–69, 2003

    2003

  36. [44]

    nanobind: Tiny and Efficient C++/Python Bindings,

    W. Jakob, “nanobind: Tiny and Efficient C++/Python Bindings,” 2022

    2022

  37. [45]

    Robot Oper- ating System 2: Design, Architecture, and Uses in the Wild,

    S. Macenski, T. Foote, B. Gerkey, C. Lalancette, and W. Woodall, “Robot Oper- ating System 2: Design, Architecture, and Uses in the Wild,”Science Robotics, vol. 7, no. 66, p. eabm6074, 2022

    2022

  38. [46]

    Botsch, L

    M. Botsch, L. Kobbelt, M. Pauly, P. Alliez, and B. Levy,Polygon Mesh Process- ing. CRC Press, Oct. 2010

    2010

  39. [47]

    Neural marching cubes,

    Z. Chen and H. Zhang, “Neural marching cubes,”ACM Trans. Graph., vol. 40, no. 6, pp. 251:1–251:15, Dec. 2021

    2021

  40. [48]

    Marching cubes: a high resolution 3D surface construction algorithm,

    W. E. Lorensen and H. E. Cline, “Marching cubes: a high resolution 3D surface construction algorithm,” inSeminal graphics: pioneering efforts that shaped the field, Volume 1. New York, NY, USA: Association for Computing Machinery, Jul. 1998, vol. Volume 1, pp. 347–353

    1998

  41. [49]

    A survey of the marching cubes algorithm,

    T. S. Newman and H. Yi, “A survey of the marching cubes algorithm,”Com- puters & Graphics, vol. 30, no. 5, pp. 854–879, Oct. 2006

    2006

  42. [50]

    Dual marching cubes,

    G. Nielson, “Dual marching cubes,” inIEEE Visualization 2004, Oct. 2004, pp. 489–496

    2004

  43. [51]

    Meshing of Surfaces,

    J.-D. Boissonnat, D. Cohen-Steiner, B. Mourrain, G. Rote, and G. Vegter, “Meshing of Surfaces,” inEffective Computational Geometry for Curves and Surfaces, J.-D. Boissonnat and M. Teillaud, Eds. Berlin, Heidelberg: Springer, 2006, pp. 181–229

    2006

  44. [52]

    Delaunay stability via pertur- bations,

    J.-D. Boissonnat, R. Dyer, and A. Ghosh, “Delaunay stability via pertur- bations,”International Journal of Computational Geometry & Applications, vol. 24, no. 02, pp. 125–152, Jun. 2014

    2014

  45. [53]

    Anisotropic Delaunay Mesh Generation,

    J.-D. Boissonnat, C. Wormser, and M. Yvinec, “Anisotropic Delaunay Mesh Generation,”SIAM Journal on Computing, vol. 44, no. 2, pp. 467–512, Jan. 2015

    2015

  46. [54]

    Quadtrees, Octrees, and Orthtrees,

    J. Campolattaro, S. Giraudot, C. Portaneri, T. Zhao, and P. Alliez, “Quadtrees, Octrees, and Orthtrees,” inCGAL User and Reference Manual, 6th ed. CGAL Editorial Board, 2024. [Online]. Available: https://doc.cgal.org/6.0.1/Manual/packages.html#PkgOrthtree

    2024

  47. [55]

    Geometric modeling using octree encoding,

    D. Meagher, “Geometric modeling using octree encoding,”Computer Graphics and Image Processing, vol. 19, no. 2, pp. 129–147, Jun. 1982

    1982

  48. [56]

    An Overview of Quadtrees, Octrees, and Related Hierarchical Data Structures,

    H. Samet, “An Overview of Quadtrees, Octrees, and Related Hierarchical Data Structures,” inTheoretical Foundations of Computer Graphics and CAD, R. A. Earnshaw, Ed. Berlin, Heidelberg: Springer, 1988, pp. 51–68

    1988

  49. [57]

    Rods and Rings: Soft Subdivision Planner for Rˆ3 x Sˆ2,

    C.-H. Hsu, Y.-J. Chiang, and C. Yap, “Rods and Rings: Soft Subdivision Planner for Rˆ3 x Sˆ2,” Jun. 2019, arXiv:1903.09416 [cs]

    2019 arXiv

  50. [58]

    On soft predicates in subdivision motion planning,

    C. Wang, Y.-J. Chiang, and C. Yap, “On soft predicates in subdivision motion planning,” inProceedings of the twenty-ninth annual symposium on Computa- tional geometry, ser. SoCG ’13. New York, NY, USA: Association for Computing Machinery, Jun. 2013, pp. 349–358

    2013

  51. [59]

    Soft Subdivision Search in Motion Planning, II: Axiomatics,

    C. K. Yap, “Soft Subdivision Search in Motion Planning, II: Axiomatics,” in Frontiers in Algorithmics, J. Wang and C. Yap, Eds. Cham: Springer Interna- tional Publishing, 2015, pp. 7–22

    2015

  52. [60]

    A Note on the Time Complexity of Using Sub- division Methods for the Approximation of Fibers,

    M. M. Bilevich and D. Halperin, “A Note on the Time Complexity of Using Sub- division Methods for the Approximation of Fibers,” Mar. 2025, arXiv:2503.01626 [cs]

    2025 arXiv

  53. [61]

    Planning Algorithms,

    S. M. LaValle, “Planning Algorithms,” inPlanning Algorithms. Cambridge University Press, May 2006, pp. 105–152

    2006

  54. [62]

    Bayesian filtering for location estimation,

    V. Fox, J. Hightower, L. Liao, D. Schulz, and G. Borriello, “Bayesian filtering for location estimation,”IEEE Pervasive Computing, vol. 2, no. 3, pp. 24–33, Jul. 2003

    2003

  55. [63]

    Robust Monte Carlo Localiza- tion for Mobile Robots,

    S. Thrun, D. Fox, W. Burgard, and F. Dellaert, “Robust Monte Carlo Localiza- tion for Mobile Robots,”Artificial Intelligence, vol. 128, no. 1, pp. 99–141, May 2001

    2001

  56. [64]

    Evans,Measure Theory and Fine Properties of Functions

    L. Evans,Measure Theory and Fine Properties of Functions. New York: Rout- ledge, Apr. 2018

    2018

  57. [65]

    Monte Carlo localization for mobile robots,

    F. Dellaert, D. Fox, W. Burgard, and S. Thrun, “Monte Carlo localization for mobile robots,” inProceedings 1999 IEEE International Conference on Robotics and Automation (Cat. No.99CH36288C), vol. 2, May 1999, pp. 1322–1328 vol.2, iSSN: 1050-4729

    1999

  58. [66]

    Random coverings in several dimensions,

    S. Janson, “Random coverings in several dimensions,”Acta Mathematica, vol. 156, no. none, pp. 83–118, Jan. 1986, publisher: Institut Mittag-Leffler

    1986

  59. [67]

    OpenMP Application Program Interface Version 3.0,

    OpenMP Architecture Review Board, “OpenMP Application Program Interface Version 3.0,” May 2008. [Online]. Available: http://www.openmp.org/ mp-documents/spec30.pdf

    2008

  60. [68]

    2D and 3D Fast Intersection and Distance Computation,

    P. Alliez, S. Tayeb, and C. Wormser, “2D and 3D Fast Intersection and Distance Computation,” inCGAL User and Reference Manual, 6th ed. CGAL Editorial Board, 2024. [Online]. Available: https://doc.cgal.org/6.0.1/Manual/ packages.html#PkgAABBTree

    2024

  61. [69]

    SLAM Toolbox: SLAM for the dynamic world,

    S. Macenski and I. Jambrecic, “SLAM Toolbox: SLAM for the dynamic world,” Journal of Open Source Software, vol. 6, no. 61, p. 2783, May 2021

    2021

  62. [70]

    The Marathon 2: A Navigation System,

    S. Macenski, F. Martin, R. White, and J. Gin´ es Clavero, “The Marathon 2: A Navigation System,” in2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2020

    2020

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.