pith. machine review for the scientific record. sign in

arxiv: 2605.06100 · v1 · submitted 2026-05-07 · 📡 eess.SP · cs.AI· cs.LG· cs.RO

Recognition: unknown

CredibleDFGO: Differentiable Factor Graph Optimization with Credibility Supervision

Liang Qian , Penggao Yan , Penghui Xu , Li-Ta Hsu
Authors on Pith no claims yet

Pith reviewed 2026-05-08 07:02 UTC · model grok-4.3

classification 📡 eess.SP cs.AIcs.LGcs.RO
keywords GNSS positioningfactor graph optimizationuncertainty estimationdifferentiable solverurban navigationcovariance credibilityproper scoring rulessatellite weighting
0
0 comments X

The pith

Training GNSS factor graphs end-to-end with scoring rules on output covariance produces more credible uncertainty estimates while also sharpening position fixes in urban canyons.

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

The paper shows that existing differentiable factor graph methods improve GNSS position estimates but leave the reported covariance unreliable because they only optimize the mean. CredibleDFGO adds explicit supervision of the full predictive distribution by letting a weighting network assign per-satellite reliability, passing those weights through a differentiable Gauss-Newton solver, and then scoring the resulting East-North covariance with negative log-likelihood and energy score. This produces posterior covariances whose credibility improves consistently on urban test scenes, with the largest accuracy gains appearing in medium- and harsh-urban conditions. The approach therefore turns the covariance from a side product into a trainable target whose shape and scale better match actual positioning errors.

Core claim

By making covariance credibility an explicit training target, the framework maps per-satellite weights produced by a Weighting Generation Network through a differentiable Gauss-Newton solver to both a position estimate and a posterior covariance, then supervises the East-North predictive distribution end-to-end with proper scoring rules (NLL, ES, and their combination). On three UrbanNav scenes this yields consistent gains in uncertainty credibility; positioning accuracy also improves, with the harsh-urban Mong Kok scene showing mean horizontal error reduced from 13.77 m to 11.68 m, NLL from 40.63 to 6.59, and ES from 12.31 to 9.05, linked to better axis-wise consistency and more credible 2D

What carries the argument

Weighting Generation Network feeding per-satellite weights into a differentiable Gauss-Newton solver whose posterior covariance is directly supervised by proper scoring rules on the East-North predictive distribution.

If this is right

  • Positioning accuracy improves on medium-urban and harsh-urban scenes while uncertainty credibility gains appear across all three tested environments.
  • On the Mong Kok harsh-urban scene the mean horizontal error drops from 13.77 m to 11.68 m and the 95th-percentile error also decreases.
  • The posterior covariance exhibits better axis-wise consistency and produces more credible local ellipses after satellite-level reweighting.
  • The combination of NLL and energy score yields the largest joint improvement in both accuracy and uncertainty metrics.

Where Pith is reading between the lines

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

  • If the credibility improvement generalizes, downstream planners could use the reported covariance directly for risk-aware routing instead of inflating safety margins.
  • The same end-to-end credibility supervision could be applied to other factor-graph sensors such as IMU or camera measurements to produce joint uncertainty estimates.
  • Training on a broader set of urban and suburban traces might further close the remaining gap between reported and empirical error distributions.

Load-bearing premise

That end-to-end supervision of the output covariance with NLL and energy score will produce a posterior whose credibility and shape generalize beyond the training scenes instead of merely fitting the chosen scoring rules.

What would settle it

Measuring whether the reported covariance ellipses contain the true position at the nominal rates when the trained model is evaluated on entirely new urban GNSS traces collected with different receiver hardware or satellite constellations.

read the original abstract

Global navigation satellite system (GNSS) positioning is widely used for urban navigation, but the covariance reported by the GNSS solver is often unreliable in urban canyons. Existing differentiable factor graph optimization (DFGO) methods already learn measurement weighting through the solver, but they still use position-only objectives. As a result, the mean estimate may improve while the reported covariance remains too small, too large, or wrong in shape. In this work, we propose CredibleDFGO (CDFGO), a differentiable GNSS factor graph framework that makes covariance credibility an explicit training target. The Weighting Generation Network (WGN) predicts per-satellite reliability weights. The differentiable Gauss--Newton solver maps these weights to a position estimate and posterior covariance, and proper scoring rules supervise the East--North predictive distribution end-to-end. We study negative log-likelihood (NLL), Energy Score (ES), and their combination. Results on three UrbanNav test scenes show consistent gains in uncertainty credibility. Positioning accuracy also improves on the medium-urban and harsh-urban scenes, and the mean horizontal error and 95th-percentile error improve on the deep-urban scene. On the harsh-urban Mong Kok (MK) scene, CDFGO-Combined reduces the mean horizontal error from 13.77\,m to 11.68\,m, reduces NLL from 40.63 to 6.59, and reduces ES from 12.31 to 9.05. The case studies link the MK improvement to better axis-wise consistency, more credible local covariance ellipses, and satellite-level reweighting.

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

3 major / 2 minor

Summary. The paper proposes CredibleDFGO (CDFGO), a differentiable factor graph optimization framework for GNSS urban positioning. A Weighting Generation Network (WGN) predicts per-satellite reliability weights that are fed into a differentiable Gauss-Newton solver; the resulting position estimate and posterior covariance are supervised end-to-end by proper scoring rules (negative log-likelihood, Energy Score, and their combination) applied to the East-North predictive distribution. Experiments on three UrbanNav scenes report reductions in mean horizontal error, NLL, and ES, with case studies attributing gains to improved axis-wise consistency and satellite reweighting.

Significance. If the reported covariance credibility generalizes, the work would be a meaningful advance for reliable uncertainty quantification in GNSS navigation. Directly optimizing proper scoring rules on the solver output addresses the common failure mode in existing DFGO methods where mean accuracy improves but reported covariances remain miscalibrated. The explicit use of NLL and ES as training targets, together with the differentiable solver, provides a clean mechanism that could be adopted in other factor-graph pipelines.

major comments (3)
  1. [Abstract and experimental results] The central claim that credibility supervision produces covariances whose calibration and shape generalize beyond the training scenes is load-bearing, yet the evaluation is confined to three UrbanNav test scenes whose multipath/NLOS statistics may be correlated with the training data. No cross-dataset or leave-one-scene-out experiments are described to test whether the learned WGN weights and Hessian-derived covariances remain credible on unseen urban layouts.
  2. [Abstract] The abstract states that NLL drops from 40.63 to 6.59 and ES from 12.31 to 9.05 on the MK scene, but supplies no implementation details on how the East-North marginal is extracted from the full posterior covariance, how the scoring rules are normalized, or the precise form of the combined loss. Without these, it is impossible to verify that the reported gains are not artifacts of the particular scoring-rule implementation or of scene-specific fitting.
  3. [Experimental evaluation] The manuscript does not report baseline comparisons against the position-only DFGO methods mentioned in the introduction, nor does it include ablation studies isolating the contribution of NLL versus ES versus the combined objective. This omission makes it difficult to attribute the observed accuracy and credibility gains specifically to the credibility supervision rather than to other modeling choices.
minor comments (2)
  1. [Method] Notation for the posterior covariance (e.g., how the 2-D East-North marginal is obtained from the 3-D or higher-dimensional Hessian) should be introduced explicitly with an equation reference.
  2. [Case studies] Figure captions for the covariance ellipses should state the confidence level (e.g., 95 %) and whether the ellipses are marginal or joint.

Simulated Author's Rebuttal

3 responses · 0 unresolved

Thank you for your constructive review. We address each of the three major comments below, clarifying our approach and committing to revisions that enhance the manuscript's rigor and transparency.

read point-by-point responses

  1. Referee: [Abstract and experimental results] The central claim that credibility supervision produces covariances whose calibration and shape generalize beyond the training scenes is load-bearing, yet the evaluation is confined to three UrbanNav test scenes whose multipath/NLOS statistics may be correlated with the training data. No cross-dataset or leave-one-scene-out experiments are described to test whether the learned WGN weights and Hessian-derived covariances remain credible on unseen urban layouts.

    Authors: We agree that stronger evidence for generalization is needed. The UrbanNav scenes were selected for their diversity in urban density, and internal train/test splits were used per scene. To address the referee's concern, we will incorporate leave-one-scene-out experiments in the revision, training the model on two scenes and evaluating on the held-out scene to verify that credibility gains persist on unseen layouts. We will also add a discussion of potential limitations in generalizing to entirely new datasets. This revision will be marked as 'partial' since full cross-dataset validation on external GNSS datasets is not feasible without additional resources. revision: partial

  2. Referee: [Abstract] The abstract states that NLL drops from 40.63 to 6.59 and ES from 12.31 to 9.05 on the MK scene, but supplies no implementation details on how the East-North marginal is extracted from the full posterior covariance, how the scoring rules are normalized, or the precise form of the combined loss. Without these, it is impossible to verify that the reported gains are not artifacts of the particular scoring-rule implementation or of scene-specific fitting.

    Authors: We regret the omission of these details. The East-North marginal is the 2x2 covariance submatrix for the East and North states from the full posterior. Scoring rules are applied using standard formulas for 2D Gaussians with no extra normalization. The combined loss is L = NLL + λ ES where λ is a hyperparameter tuned on validation data. We will update the abstract with a concise description and add precise equations and implementation details to the methods section in the revised manuscript. revision: yes

  3. Referee: [Experimental evaluation] The manuscript does not report baseline comparisons against the position-only DFGO methods mentioned in the introduction, nor does it include ablation studies isolating the contribution of NLL versus ES versus the combined objective. This omission makes it difficult to attribute the observed accuracy and credibility gains specifically to the credibility supervision rather than to other modeling choices.

    Authors: This is a valid criticism. We will add comparisons to a position-only DFGO baseline (trained with position error loss only) using the same network and solver. We will also provide ablation results for NLL-only, ES-only, and combined supervision to clearly isolate the effect of credibility supervision. These will be included in the experimental section of the revised paper. revision: yes

Circularity Check

0 steps flagged

No circularity: end-to-end supervision via proper scoring rules is independent of the covariance parameterization

full rationale

The derivation proceeds as: WGN produces per-satellite weights; differentiable Gauss-Newton solver maps weights to position estimate and posterior covariance (via Hessian); NLL/ES losses are applied to the resulting East-North Gaussian predictive distribution against ground-truth positions. Proper scoring rules are standard, externally defined objectives whose optimum occurs when the predictive distribution matches the empirical error distribution; they do not embed the solver's covariance formula by construction. No equation equates the learned covariance to a fitted parameter or renames an input as output. No load-bearing self-citation chain is present; the framework builds on prior DFGO work but adds an independent credibility objective. The reported gains are therefore not forced by definition or by re-labeling of training targets.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, invented entities, or non-standard axioms are named. The framework rests on the standard assumption that a differentiable Gauss-Newton solver produces a usable posterior covariance and that proper scoring rules can calibrate it.

axioms (2)
  • domain assumption Differentiable Gauss-Newton solver maps per-satellite weights to a posterior covariance whose credibility can be supervised by NLL and ES
    Invoked when the paper states that the solver produces the covariance used for scoring-rule supervision.
  • domain assumption Proper scoring rules applied to the East-North predictive distribution produce credible full covariance estimates
    Central to the training target described in the abstract.

pith-pipeline@v0.9.0 · 5601 in / 1477 out tokens · 85210 ms · 2026-05-08T07:02:57.588691+00:00 · methodology

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

53 extracted references · 2 canonical work pages

  1. [1]

    Zico , title =

    Amos, Brandon and Kolter, J. Zico , title =. Proceedings of the 34th International Conference on Machine Learning (ICML) , series =. 2017 , publisher =

    2017

  2. [2]

    Proceedings of the 2020 International Technical Meeting of The Institute of Navigation , pages =

    Garcia Crespillo, Omar and Andreetti, Alice and Grosch, Anja , title =. Proceedings of the 2020 International Technical Meeting of The Institute of Navigation , pages =. 2020 , organization=

    2020

  3. [3]

    Foundations and Trends in Robotics , volume =

    Dellaert, Frank and Kaess, Michael , title =. Foundations and Trends in Robotics , volume =. 2017 , doi =

    2017

  4. [4]

    Proceedings of the 33rd International Conference on Machine Learning (ICML) , series =

    Gal, Yarin and Ghahramani, Zoubin , title =. Proceedings of the 33rd International Conference on Machine Learning (ICML) , series =. 2016 , publisher =

    2016

  5. [5]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Gallon, Elisa and Joerger, Mathieu and Pervan, Boris , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2022 , doi =

    2022

  6. [6]

    , title =

    Gneiting, Tilmann and Raftery, Adrian E. , title =. Journal of the American Statistical Association , volume =. 2007 , doi =

    2007

  7. [7]

    and Grimit, Eric P

    Gneiting, Tilmann and Stanberry, Larissa I. and Grimit, Eric P. and Held, Leonhard and Johnson, Nicholas A. , title =. TEST , volume =. 2008 , doi =

    2008

  8. [8]

    Journal of the American Statistical Association , volume =

    Gneiting, Tilmann , title =. Journal of the American Statistical Association , volume =. 2011 , doi =

    2011

  9. [9]

    Proceedings of the 34th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2021) , pages =

    Hsu, Li-Ta and Kubo, Nobuaki and Wen, Weisong and Chen, Wu and Liu, Zhizhao and Suzuki, Taro and Meguro, Junichi , title =. Proceedings of the 34th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2021) , pages =. 2021 , organization=

    2021

  10. [10]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Hsu, Li-Ta and Huang, Feng and Ng, Hoi-Fung and Zhang, Guohao and Zhong, Yihan and Bai, Xiwei and Wen, Weisong , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2023 , doi =

    2023

  11. [11]

    IEEE Transactions on Intelligent Transportation Systems , volume =

    Hu, Runzhi and Xu, Penghui and Zhong, Yihan and Wen, Weisong , title =. IEEE Transactions on Intelligent Transportation Systems , volume =. 2025 , doi =

    2025

  12. [12]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Joerger, Mathieu and Pervan, Boris , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2020 , doi =

    2020

  13. [13]

    and Dellaert, Frank , title =

    Kaess, Michael and Johannsson, Hordur and Roberts, Richard and Ila, Viorela and Leonard, John J. and Dellaert, Frank , title =. The International Journal of Robotics Research , volume =. 2012 , doi =

    2012

  14. [14]

    and Gupta, Shubh and Shetty, Akshay and Gao, Grace , title =

    Kanhere, Ashwin V. and Gupta, Shubh and Shetty, Akshay and Gao, Grace , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2022 , doi =

    2022

  15. [15]

    and Hegarty, Christopher , title =

    Kaplan, Elliott D. and Hegarty, Christopher , title =. 2017 , publisher =

    2017

  16. [16]

    Auto-Encoding Variational Bayes

    Kingma, Diederik P. and Welling, Max , title =. International Conference on Learning Representations (ICLR) , year =. doi:10.48550/arXiv.1312.6114 , url =

    work page Pith review doi:10.48550/arxiv.1312.6114

  17. [17]

    Autonomous Robots , volume =

    Kloss, Alina and Martius, Georg and Bohg, Jeannette , title =. Autonomous Robots , volume =. 2021 , doi =

    2021

  18. [18]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Knowles, Derek and Gao, Grace , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2023 , doi =

    2023

  19. [19]

    and Frey, Brendan J

    Kschischang, Frank R. and Frey, Brendan J. and Loeliger, Hans-Andrea , title =. IEEE Transactions on Information Theory , volume =. 2001 , doi =

    2001

  20. [20]

    Sensors , volume =

    Levi, Dan and Gispan, Liran and Giladi, Niv and Fetaya, Ethan , title =. Sensors , volume =. 2022 , doi =

    2022

  21. [21]

    Rong and Zhao, Zhanlue and Li, Xiao-Bai , title =

    Li, X. Rong and Zhao, Zhanlue and Li, Xiao-Bai , title =. IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans , volume =. 2012 , doi =

    2012

  22. [22]

    Satellite Navigation , volume =

    Li, Lintong and Elhajj, Mireille and Feng, Yuxiang and Ochieng, Washington Yotto , title =. Satellite Navigation , volume =. 2023 , doi =

    2023

  23. [23]

    and Ba, Jimmy , title =

    Kingma, Diederik P. and Ba, Jimmy , title =. Proceedings of the 3rd International Conference on Learning Representations (ICLR) , year =

  24. [24]

    and Brunner, F

    Hartinger, H. and Brunner, F. K. , title =. GPS Solutions , volume =. 1999 , doi =

    1999

  25. [25]

    Measurement Science and Technology , volume =

    Realini, Eugenio and Reguzzoni, Mirko , title =. Measurement Science and Technology , volume =. 2013 , doi =

    2013

  26. [26]

    Sensors , volume =

    Medina, Daniel and Li, Haoqing and Vila-Valls, Jordi and Closas, Pau , title =. Sensors , volume =. 2019 , doi =

    2019

  27. [27]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Mohanty, Adyasha and Gao, Grace , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2023 , doi =

    2023

  28. [28]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Mohanty, Adyasha and Gao, Grace , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2024 , doi =

    2024

  29. [29]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Nagai, Kana and Spenko, Matthew and Henderson, Ron and Pervan, Boris , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2024 , doi =

    2024

  30. [30]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Ng, Hoi-Fung and Zhang, Guohao and Luo, Yiran and Hsu, Li-Ta , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2021 , doi =

    2021

  31. [31]

    Stochastic Backpropagation and Approximate Inference in Deep Generative Models

    Rezende, Danilo Jimenez and Mohamed, Shakir and Wierstra, Daan , title =. Proceedings of the 31st International Conference on Machine Learning , series =. 2014 , publisher =. doi:10.48550/arXiv.1401.4082 , url =

    work page Pith review doi:10.48550/arxiv.1401.4082 2014

  32. [32]

    , title =

    Scheuerer, Michael and Hamill, Thomas M. , title =. Monthly Weather Review , volume =. 2015 , doi =

    2015

  33. [33]

    , title =

    Smolyakov, Ivan and Rezaee, Mohammad and Langley, Richard B. , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2020 , doi =

    2020

  34. [34]

    IEEE Internet of Things Journal , volume =

    Chen, Wu and Sun, Rui and Fu, Linxia and Cheng, Qi and Chiang, Kai Wei , title =. IEEE Internet of Things Journal , volume =. 2023 , doi =

    2023

  35. [35]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Taylor, Clark and Gross, Jason , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2024 , doi =

    2024

  36. [36]

    Springer Handbook of Global Navigation Satellite Systems , year =

  37. [37]

    Pineda, Luis and Fan, Taosha and Monge, Maurizio and Venkataraman, Shobha and Sodhi, Paloma and Chen, Ricky T. Q. and Ortiz, Joseph and DeTone, Daniel and Wang, Austin and Anderson, Stuart and Dong, Jing and Amos, Brandon and Mukadam, Mustafa , title =. Advances in Neural Information Processing Systems (. 2022 , doi =

    2022

  38. [38]

    and Kaiser, Lukasz and Polosukhin, Illia , title =

    Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Jakob and Jones, Llion and Gomez, Aidan N. and Kaiser, Lukasz and Polosukhin, Illia , title =. Advances in Neural Information Processing Systems (. 2017 , doi =

    2017

  39. [39]

    and Ziebart, Marek K

    Wang, Lei and Groves, Paul D. and Ziebart, Marek K. , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2013 , doi =

    2013

  40. [40]

    2020 IEEE/ION Position, Location and Navigation Symposium (PLANS) , pages =

    Wen, Weisong and Bai, Xiwei and Hsu, Li-Ta and Pfeifer, Tim , title =. 2020 IEEE/ION Position, Location and Navigation Symposium (PLANS) , pages =. 2020 , organization=

    2020

  41. [41]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Wen, Weisong and Pfeifer, Tim and Bai, Xiwei and Hsu, Li-Ta , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2021 , doi =

    2021

  42. [42]

    2021 IEEE International Conference on Robotics and Automation (ICRA) , pages =

    Wen, Weisong and Hsu, Li-Ta , title =. 2021 IEEE International Conference on Robotics and Automation (ICRA) , pages =. 2021 , organization=

    2021

  43. [43]

    IEEE Transactions on Vehicular Technology , volume =

    Wen, Weisong and Zhang, Guohao and Hsu, Li-Ta , title =. IEEE Transactions on Vehicular Technology , volume =. 2022 , doi =

    2022

  44. [44]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Xia, Xiao and Wen, Weisong and Hsu, Li-Ta , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2024 , doi =

    2024

  45. [45]

    Proceedings of the 36th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2023) , pages =

    Xu, Penghui and Ng, Hoi-Fung and Zhong, Yihan and Zhang, Guohao and Wen, Weisong and Yang, Bo and Hsu, Li-Ta , title =. Proceedings of the 36th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2023) , pages =. 2023 , doi =

    2023

  46. [46]

    Proceedings of the 37th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2024) , pages =

    Xu, Penghui and Hsu, Li-Ta , title =. Proceedings of the 37th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2024) , pages =. 2024 , doi =

    2024

  47. [47]

    IEEE Transactions on Intelligent Vehicles , volume =

    Yan, Penggao and Xia, Xiao and Brizzi, Michele and Wen, Weisong and Hsu, Li-Ta , title =. IEEE Transactions on Intelligent Vehicles , volume =. 2025 , doi =

    2025

  48. [48]

    and Kloss, Alina and Martin-Martin, Roberto and Bohg, Jeannette , title =

    Yi, Brent and Lee, Michelle A. and Kloss, Alina and Martin-Martin, Roberto and Bohg, Jeannette , title =. 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages =. 2021 , publisher =

    2021

  49. [49]

    IEEE Sensors Journal , volume =

    Zhang, Guohao and Xu, Penghui and Xu, Haosheng and Hsu, Li-Ta , title =. IEEE Sensors Journal , volume =. 2021 , doi =

    2021

  50. [50]

    NAVIGATION: Journal of the Institute of Navigation , volume =

    Zheng, Shaolong and Zeng, Kungan and Li, Zhenni and Wang, Qianming and Xie, Kan and Liu, Ming and Xie, Shengli , title =. NAVIGATION: Journal of the Institute of Navigation , volume =. 2024 , doi =

    2024

  51. [51]

    2025 , eprint =

    Yan, Penggao and Zhan, Xingqun and Sun, Rui and Hsu, Li-Ta , title =. 2025 , eprint =

    2025

  52. [52]

    IEEE Sensors Journal , volume =

    Yan, Penggao and Hu, Yingjie and Wen, Weisong and Hsu, Li-Ta , title =. IEEE Sensors Journal , volume =. 2025 , month = feb, doi =

    2025

  53. [53]

    2026 , eprint =

    Li, Zhengdao and Yan, Penggao and Hsu, Li-Ta , title =. 2026 , eprint =

    2026