Pith. sign in

REVIEW 3 major objections 6 minor 52 references

RaSCL: Radar to Satellite Crossview Localization

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A single ground radar scan can be matched to satellite imagery to localize a robot without GNSS, achieving meter-level accuracy in unseen areas.

desk verdict Solid radar-to-satellite localization system, but Table I's training leakage means the headline comparative claim rests on train-on-test evidence. read the letter →

arxiv 2504.15899 v1 pith:E5W3X6AH submitted 2025-04-22 cs.RO cs.CV

Blerim Abdullai , Tony Wang , Xinyuan Qiao , Florian Shkurti , Timothy D. Barfoot This is my paper · ORCID
classification cs.ROcs.CV
keywords GNSS-freelocalizationradarodometrycrossviewsatelliteimageryiterativeclosestpointoccupancypredictionfactorgraphoptimizationunmannedsurfacevessel
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

The paper claims that a robot can localize itself globally, without GNSS, by matching a single ground radar scan to publicly available overhead RGB imagery. It converts the satellite image into a point cloud with a learned occupancy network, extracts salient points from the radar with k-strongest filtering, and aligns the two with ICP. These overhead registration factors are then jointly optimized with scan-to-scan radar odometry in a fixed-lag smoother. On three datasets (urban driving, suburban driving, and a marine vessel) the method reports meter-level RMSE in unseen test areas, and outperforms prior radar-to-overhead methods on the Boreas dataset while remaining competitive on Oxford.

What carries the argument

The central objects are the learned occupancy point cloud and the ICP registration between it and the k-strongest radar points. The occupancy network is an Attention U-Net that takes the concatenation of RGB satellite imagery and Google Roadmap imagery, is trained with a binary cross-entropy plus Dice loss against a lidar-derived occupancy label masked by a certainty mask, and is turned into a point cloud at inference by ray tracing from the image center. The k-strongest filter keeps the k highest-intensity points per radar azimuth ($k=9$ for satellite registration, $k=5$ for odometry). Point-to-point ICP with a $4.33\,\mathrm{m}$ trim distance aligns the radar points to the predicted satellite point cloud; the resulting relative pose becomes a unary factor, and ICP between consecutive radar scans becomes a binary odometry factor, jointly optimized in a fixed-lag smoother with a georeferenced initial guess.

What would settle it

Take a sequence in an area with dense overhead foliage or recent construction; if the occupancy network predicts vegetation or outdated buildings as occupied while the radar sees through them or sees new structures, the point clouds will disagree. A concrete check: shift the satellite imagery by 10 m in a test area and measure whether ICP still converges; if it does not, the method's reliance on a close initial guess is confirmed.

Watch

Extended reading notes

Core claim

On its own terms, the discovery is that the cross-modal registration problem between ground radar and overhead RGB imagery can be reduced to a point-cloud alignment problem, provided the overhead image is first converted into a learned occupancy representation supervised by lidar. Rather than learning a direct radar-to-image feature matcher, the paper uses an Attention U-Net to predict occupancy from the satellite image plus roadmap image, ray-traces this occupancy into a point cloud, extracts radar points via k-strongest filtering, and registers the two clouds with point-to-point ICP using a trim distance. This per-scan registration is then treated as a unary factor in a factor graph, alongside binary radar odometry factors, giving a complete GNSS-free localization solution that works in previously unseen areas and in a marine environment where road-segment priors do not exist.

Load-bearing premise

The central assumption is that the learned occupancy point cloud derived from satellite imagery has the same salient geometry as the k-strongest radar points, so that ICP converges to the true pose from the current state estimate.

Editorial extensions

If this is right

  • A robot equipped only with radar and an internet connection to fetch satellite imagery can maintain global localization in unmapped areas, without GPS or a prior lidar map.
  • Because the method does not use lane lines or building footprints, it extends to off-road and marine settings where road-based priors are absent or outdated.
  • The registration is robust to dynamic occluders and temporary poor imagery because low-fitness ICP matches (below $\tau_{\mathrm{fit}}=0.6$) are discarded from the factor graph.
  • The learned occupancy representation can be precomputed over a survey area, removing the GPU requirement at deployment and allowing human correction of incorrect predictions.

Reading between the lines

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

  • If the occupancy network can be trained on one geographic region and generalize, a natural extension is to test zero-shot transfer to cities not in the training split; the paper's cross-dataset evaluation (Boreas/Oxford/Boat) leaves open how far the learned geometry transfers.
  • Because the method relies on a georeferenced initial guess and fetches imagery using the latest state estimate, a failure mode is unbounded drift if the initial guess lies outside the ICP convergence basin; adding a global place-recognition step could make the system truly global.
  • The reported Oxford errors (3.5–4.5 m) suggest that dense urban scenes with tall buildings and orthorectification artifacts are the current bottleneck; a submap-based registration or explicit handling of moving vehicles could close much of this gap.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper presents RaSCL, a GNSS-free localization system that registers a ground radar scan to overhead RGB satellite imagery. The method converts satellite imagery into a learned occupancy point cloud via an Attention U-Net, extracts radar points using k-strongest filtering, computes point-to-point ICP registrations between radar and satellite-derived point clouds, and jointly optimizes these overhead registration factors with radar odometry factors in a fixed-lag smoother. The authors evaluate on the Boreas and Oxford Radar RobotCar datasets and on a new USV dataset, reporting RMSEs of 1.3 to 1.6 m on Boreas, 3.5 to 4.5 m on Oxford, and 3.5 to 4.8 m on the Boat dataset. They claim to outperform previous radar-to-overhead localization methods on Boreas while remaining competitive on Oxford, and to localize in previously unseen terrain using only radar, public imagery, and a single georeferenced initial guess.

Significance. If the claims are substantiated, RaSCL is a practical and useful contribution: it avoids explicit road and lane priors, works with a single georeferenced initial guess, and demonstrates the integration of learned satellite-to-occupancy prediction with k-strongest radar extraction and ICP within a factor-graph smoother. The evaluation on a USV platform and the discussion of failure modes add value. The paper also provides a clear system description and a parameter table. However, the central comparative claim is currently weakened by the evaluation protocol in Table I, whose footnote discloses that the occupancy predictor was trained on a portion of the satellite images within the evaluated sequence, while the baselines receive no such exposure. The unseen-area results in Table II report only the proposed method, so the 'outperforms previous methods' claim is not yet supported by a fair comparison. The underlying method may well be sound, but the evidence as presented does not justify the comparative claim.

major comments (3)
  1. [Section IV-B, Table I] The asterisk in Table I states that the occupancy predictor has been trained on a portion of the satellite images within the sequence. Because the comparison methods [11] and [37] do not use learned occupancy, the whole-trajectory comparison is not on unseen imagery for RaSCL but is for the baselines. The reported Boreas margins (1.3 to 1.6 m versus 3.0 to 4.9 m for Li [37] and Hong [11]) are therefore confounded by possible memorization of the evaluation area. Please provide baseline results on the same unseen test-area split used in Table II, or re-run the full-trajectory evaluation under a protocol in which no training image lies on any evaluated trajectory.
  2. [Section IV-C and abstract/contributions] The sentence 'Our evaluation shows that our method outperforms previous radar-to-overhead imagery methods on Boreas while remaining competitive on the challenging Oxford dataset' is not supported for unseen areas because Table II reports only the proposed method. The 'unseen terrain' capability claim (contribution 2) can be supported by Table II alone, but the comparative 'outperforms' claim needs baseline numbers on the red test-area split. Please either add those numbers or qualify the claim to state that the comparison is only on the full trajectories with the leakage caveat made explicit.
  3. [Section III-E and Figure 2] The method fetches satellite imagery using the most recent state estimate and requires a single georeferenced initial guess; it does not perform a global search over the map. The term 'global localization' in the abstract and introduction should be qualified as 'metric localization against a global overhead map' or 'global map-based localization' to avoid implying recovery from an arbitrary initial pose, which is not demonstrated.
minor comments (6)
  1. [Section III heading] The section heading contains a typo: 'METHODOLODY' should be 'METHODOLOGY'.
  2. [Equation (1)] The superscript dagger in I_L† is used but never defined; please define it when the lidar label is introduced.
  3. [Reference [43]] The URL for the GTSAM reference has an extra closing parenthesis: 'https://github.com/borglab/gtsam)' should be 'https://github.com/borglab/gtsam'.
  4. [Section III-A] The phrase 'we keep the same m pixel' is unclear; please rephrase to specify the resolution or number of pixels used in the polar representation.
  5. [Section IV-C and Table IV] The text says the authors increased k_odom and k_sat to 20 for the Boat data, but Table IV lists default values of 5 and 9; please state the Boat-specific values explicitly in the table or caption.
  6. [Table I caption] The disclosure about training on a portion of the satellite images within the sequence is important enough that it should be discussed in the main text, not only in the table caption.

Circularity Check

1 steps flagged · score 4.0 of 10

One admitted training leak makes the head-to-head Table I comparison partially circular; the underlying factor-graph derivation is otherwise self-contained.

  1. fitted input called prediction [Table I caption, Section IV-B (Full Sequence Localization)]
    "The * indicates that the occupancy predictor has been trained on a portion of the satellite images within the sequence, and our method also uses a history of frames in a window, whereas the other methods use filtering."

    The learned occupancy extractor that feeds the satellite registration factor is fit to imagery from the very sequence whose whole-trajectory RMSE is then reported as a localization prediction in Table I. Since Table I is the only head-to-head comparison behind the paper's claim to outperform Hong [11] and Li [37], the margin on Boreas (1.3-1.6 m vs 3.0-4.9 m) is partly a test-distribution exposure effect: the network has seen the test area's satellite geometry during training, so the registration target is memorized rather than independently derived. Table II restricts to the unseen red test area but lists no baselines, so it cannot rescue the comparative outperformance claim.

full rationale

Aside from the Table I training leak, I find no derivation-level circularity. The pipeline's load-bearing pieces are external: the occupancy predictor is summarized from Tang et al. [6] (not the present authors), radar point extraction uses k-strongest filtering from CFEAR [20]/[40], ICP is Besl-McKay [18], and the fixed-lag smoother is GTSam [43]. None of these reduce to the paper's own outputs by construction. The radar-to-satellite registration factor and odometry factor are defined by independent cost functions (Eqs. 4-5) whose inputs are measurements, not the target poses. The self-citations in the paper (Boreas dataset, radar odometry references from the Barfoot group) are standard dataset/method citations and are not load-bearing assumptions smuggled in to force the result. The manuscript itself flags the * training condition, and the presence of an unseen-area Table II shows awareness of the issue, though Table II omits the baselines needed to support the outperforms-previous-methods claim on unseen terrain. Therefore the central technical derivation is self-contained; the main circularity concern is confined to the comparative evaluation, warranting a moderate score rather than a high one.

Assumptions & free parameters 11 free parameters · 4 assumptions · 0 invented entities

The method rests primarily on learned priors and hand-tuned parameters rather than new physical entities. The occupancy network is a learned map representation, not an invented physical entity; it is trained with lidar supervision from the deployment domains. The many hand-chosen thresholds and covariances mean that the reported performance is conditional on this tuning.

free parameters (11)
  • lambda (occupancy loss weight) = 0.5
    Chosen as best in experiments (Section III-A, Equation 3).
  • tau_lidar (lidar mask threshold) = 0.04
    Set for converting lidar into BEV mask M and label I_L (Section III-A).
  • tau_occ (occupancy threshold) = 0.6
    Used to ray-trace the first occupied point along each azimuth (Section III-A).
  • k_radar (K-strongest for odometry) = 5
    Chosen for best radar odometry (Section III-B).
  • k_sat (K-strongest for overhead registration) = 9
    Chosen for best radar-to-satellite registration (Section III-B).
  • delta_sat (satellite ICP trim) = 4.33 m (10 pixels)
    Best trim distance for overhead registration; ICP initialized with a 50-pixel trim for five iterations (Section III-C).
  • delta_radar (odometry ICP trim) = 4 m
    Trim distance for consecutive radar scan ICP (Section III-D).
  • tau_fit (ICP fitness threshold) = 0.6
    Filters low-quality satellite registrations before adding them to the factor graph (Section III-E, Equation 6).
  • sigma_odom_xy and sigma_odom_yaw = 0.04 m and 0.1 degrees
    Isotropic odometry covariance parameters for the factor graph (Table IV).
  • sigma_sat_xy and sigma_sat_yaw = 0.5 m and 4.5 degrees
    Satellite registration covariance parameters (Table IV).
  • sliding window W = 10 seconds
    Fixed-lag smoother window size (Section III-E).
assumptions (4)
  • domain assumption Radar k-strongest point extraction and lidar-supervised occupancy prediction capture the same salient geometry needed for ICP.
    The registration in Section III-C aligns radar points with ray-traced occupancy points; if the two modalities highlight different structures, the ICP matches are meaningless. The paper does not provide an independent geometric correspondence test beyond the final localization errors.
  • domain assumption Google Maps satellite and roadmap imagery is orthorectified and geographically consistent at the working resolution.
    All point clouds from overhead imagery are generated in image pixel coordinates and assumed to align with the metric map frame (Sections III-A and III-E). Any georeferencing error enters directly into the localization error.
  • domain assumption The provided georeferenced initial guess lies within the convergence basin of point-to-point ICP for the first satellite registration.
    Section III-E starts optimization from a WGS84 origin and heading, and Section III-C uses ICP with a 50-pixel initial trim to recover from larger errors, but there is no global search or place-recognition step. If the initial guess is outside the basin, the estimate can converge to a wrong local minimum.
  • domain assumption Lidar-derived binary labels and certainty masks are representative of radar-visible occupancy in the test areas.
    The occupancy network is trained with lidar labels (Section III-A) and applied to radar at test time; no radar-supervision or domain-adaptation step is used.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RaSCL: Radar to Satellite Crossview Localization." pith.science (2026). https://pith.science/paper/E5W3X6AH

@misc{pith2026250415899,
  author       = {Pith},
  title        = {Pith review of: RaSCL: Radar to Satellite Crossview Localization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E5W3X6AH}},
  note         = {Machine review of arXiv:2504.15899}
}
read the original abstract

GNSS is unreliable, inaccurate, and insufficient in many real-time autonomous field applications. In this work, we present a GNSS-free global localization solution that contains a method of registering imaging radar on the ground with overhead RGB imagery, with joint optimization of relative poses from odometry and global poses from our overhead registration. Previous works have used various combinations of ground sensors and overhead imagery, and different feature extraction and matching methods. These include various handcrafted and deep-learning-based methods for extracting features from overhead imagery. Our work presents insights on extracting essential features from RGB overhead images for effective global localization against overhead imagery using only ground radar and a single georeferenced initial guess. We motivate our method by evaluating it on datasets in diverse geographic conditions and robotic platforms, including on an Unmanned Surface Vessel (USV) as well as urban and suburban driving datasets.

Figures

Figures reproduced from arXiv: 2504.15899 by the authors.

Figure 1
Figure 1. Top shows our Otter USV data collection platform [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. A. Learning To Predict Points This section summarizes the methodology in [6], in which the authors present a learning method to generate a point cloud given overhead imagery. To extract a point cloud from a satellite image, we first concatenate the RGB Overhead image and the Google Roadmaps image, which together we refer to as IS and pass it as input to an Attention U-Net [39] to predict O, a learned occupancy repre… view at source ↗
Figure 2
Figure 2. Overview of localization pipeline. The red factors are binary factors formed using the solution from point-to-point [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figures from the paper (6 more)
Figure 3
Figure 3. Figure 3: Overview of training pipeline for the learned occu [PITH_FULL_IMAGE:figures/full_fig_p004_3.png]
Figure 4
Figure 4. Figure 4: Odometry drift visualization. In this comparison, [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Dataset geographic splits. Blue is training, red is test, [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 5
Figure 5. Figure 5: We use the Google Maps API to retrieve satellite [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Comparison of roadmap imagery with overhead RGB [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Localization error analysis: The top of each subfigure contains the latitude and longitude errors in meters vs time [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 40 canonical work pages

  1. [11]

    Large-scale radar localization using online public maps,

    Z. Hong, Y . Petillot, K. Zhang, S. Xu, and S. Wang, “Large-scale radar localization using online public maps,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) , 2023, pp. 3990– 3996

    work page 2023

  2. [37]

    Uav localization using autoencoded satellite images,

    M. Bianchi and T. D. Barfoot, “Uav localization using autoencoded satellite images,” IEEE Robotics and Automation Letters , vol. 6, no. 2, pp. 1761–1768, 2021

    work page 2021

  3. [10]

    Self-supervised learning for using overhead imagery as maps in outdoor range sensor localization,

    T. Y . Tang, D. D. Martini, S. Wu, and P. Newman, “Self-supervised learning for using overhead imagery as maps in outdoor range sensor localization,” The International Journal of Robotics Research , vol. 40, no. 12-14, pp. 1488–1509, 2021, pMID: 34992328

    work page 2021

  4. [1]

    Are we ready for radar to replace lidar in all-weather mapping and localization?

    K. Burnett, Y . Wu, D. J. Yoon, A. P. Schoellig, and T. D. Barfoot, “Are we ready for radar to replace lidar in all-weather mapping and localization?” IEEE Robotics and Automation Letters , vol. 7, no. 4, pp. 10 328–10 335, 2022

    work page 2022

  5. [2]

    Introduction to radar,

    M. I. Skolnik, “Introduction to radar,” Radar handbook , vol. 2, p. 21, 1962

    work page 1962

  6. [3]

    An overview of autonomous vehicles sensors and their vulnerability to weather conditions,

    J. Vargas, S. Alsweiss, O. Toker, R. Razdan, and J. Santos, “An overview of autonomous vehicles sensors and their vulnerability to weather conditions,” Sensors, vol. 21, no. 16, 2021

    work page 2021

  7. [4]

    Radar teach and repeat: Architecture and initial field testing,

    X. Qiao, A. Krawciw, S. Lilge, and T. D. Barfoot, “Radar teach and repeat: Architecture and initial field testing,” arXiv preprint arXiv:2409.10491, 2024

    arXiv 2024

  8. [5]

    Uncertainty-aware vision-based metric cross-view geolocal- ization,

    F. Fervers, S. Bullinger, C. Bodensteiner, M. Arens, and R. Stiefel- hagen, “Uncertainty-aware vision-based metric cross-view geolocal- ization,” in Proc. IEEE/CVF Conf. Comput. Vision Pattern Recognit. (CVPR), June 2023, pp. 21 621–21 631

    work page 2023

Show all 52 references
  1. [6]

    Point-based metric and topological localisation between lidar and overhead imagery,

    T. Y . Tang, D. De Martini, and P. Newman, “Point-based metric and topological localisation between lidar and overhead imagery,” Autonomous Robots , vol. 47, no. 5, pp. 595–615, Jun 2023

    2023

  2. [7]

    Continuous self-localization on aerial images using visual and lidar sensors,

    F. Fervers, S. Bullinger, C. Bodensteiner, M. Arens, and R. Stiefelha- gen, “Continuous self-localization on aerial images using visual and lidar sensors,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots Syst. (IROS) . IEEE, 2022, pp. 7028–7035

    2022

  3. [8]

    Overhead image factors for underwater sonar-based slam,

    J. McConnell, F. Chen, and B. Englot, “Overhead image factors for underwater sonar-based slam,” 2022

    2022

  4. [9]

    Rsl-net: Localising in satellite images from a radar on the ground,

    T. Y . Tang, D. D. Martini, D. Barnes, and P. Newman, “Rsl-net: Localising in satellite images from a radar on the ground,” 2020

    2020

  5. [12]

    R. J. Doviak and D. S. Zrnic, Doppler radar & weather observations . Academic press, 2014

    2014

  6. [13]

    J. P. Fitch, Synthetic aperture radar . Springer Science & Business Media, 2012

    2012

  7. [14]

    H. M. Jol, Ground penetrating radar theory and applications . else- vier, 2008

    2008

  8. [15]

    Phased-array and radar breakthroughs,

    E. Brookner, “Phased-array and radar breakthroughs,” in 2007 IEEE Radar Conference, 2007, pp. 37–42

    2007

  9. [16]

    Inverse synthetic aperture radar imag- ing: A historical perspective and state-of-the-art survey,

    R. Vehmas and N. Neuberger, “Inverse synthetic aperture radar imag- ing: A historical perspective and state-of-the-art survey,” IEEE Access, vol. 9, pp. 113 917–113 943, 2021

    2021

  10. [17]

    Lodestar: Maritime radar descriptor for semi-direct radar odometry,

    H. Jang, M. Jung, M.-H. Jeon, and A. Kim, “Lodestar: Maritime radar descriptor for semi-direct radar odometry,” IEEE Robotics and Automation Letters, vol. 9, no. 2, p. 1684–1691, Feb. 2024

    2024

  11. [18]

    A method for registration of 3-d shapes,

    P. Besl and N. D. McKay, “A method for registration of 3-d shapes,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 14, no. 2, pp. 239–256, 1992

    1992

  12. [19]

    Radar-only ego-motion estimation in dif- ficult settings via graph matching,

    S. H. Cen and P. Newman, “Radar-only ego-motion estimation in dif- ficult settings via graph matching,” in 2019 International Conference on Robotics and Automation (ICRA) , 2019, pp. 298–304

    2019

  13. [20]

    Lidar-level localization with radar? the cfear ap- proach to accurate, fast, and robust large-scale radar odometry in diverse environments,

    D. Adolfsson, M. Magnusson, A. Alhashimi, A. J. Lilienthal, and H. Andreasson, “Lidar-level localization with radar? the cfear ap- proach to accurate, fast, and robust large-scale radar odometry in diverse environments,” IEEE Transactions on robotics , vol. 39, no. 2, pp. 1476...

    2022

  14. [21]

    Under the radar: Learning to predict robust keypoints for odometry estimation and metric localisation in radar,

    D. Barnes and I. Posner, “Under the radar: Learning to predict robust keypoints for odometry estimation and metric localisation in radar,”

  15. [22]

    Radar odometry combining probabilistic estimation and unsupervised feature learning,

    K. Burnett, D. J. Yoon, A. P. Schoellig, and T. D. Barfoot, “Radar odometry combining probabilistic estimation and unsupervised feature learning,” 2021. [Online]. Available: https://arxiv.org/abs/2105.14152

    2021 arXiv

  16. [23]

    Masking by moving: Learning distraction-free radar odometry from pose information,

    D. Barnes, R. Weston, and I. Posner, “Masking by moving: Learning distraction-free radar odometry from pose information,”

  17. [24]

    Improving marine radar odometry by modeling radar resolution and exploiting additional temporal information,

    C. H. Schiller, B. Arsenali, D. Maas, and S. Maranó, “Improving marine radar odometry by modeling radar resolution and exploiting additional temporal information,” in 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , 2022, pp. 8436– 8441

    2022

  18. [25]

    Available: https://arxiv.org/abs/1909.03752

    [Online]. Available: https://arxiv.org/abs/1909.03752

    1909 arXiv

  19. [26]

    Radar slam: A robust slam system for all weather conditions,

    Z. Hong, Y . Petillot, A. Wallace, and S. Wang, “Radar slam: A robust slam system for all weather conditions,” arXiv preprint arXiv:2104.05347, 2021

    2021 arXiv

  20. [27]

    Are doppler velocity measurements useful for spinning radar odometry?

    D. Lisus, K. Burnett, D. J. Yoon, R. Poulton, J. Marshall, and T. D. Barfoot, “Are doppler velocity measurements useful for spinning radar odometry?” IEEE Robotics and Automation Letters , 2024

    2024

  21. [28]

    Driven to distraction: Self-supervised distractor learning for robust monocular visual odometry in urban environments,

    D. Barnes, W. Maddern, G. Pascoe, and I. Posner, “Driven to distraction: Self-supervised distractor learning for robust monocular visual odometry in urban environments,” in 2018 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2018, pp. 1894–1900

    2018

  22. [29]

    Radarize: Enhancing radar slam with generalizable doppler-based odometry,

    E. Sie, X. Wu, H. Guo, and D. Vasisht, “Radarize: Enhancing radar slam with generalizable doppler-based odometry,” in Proceedings of the 22nd Annual International Conference on Mobile Systems, Applications and Services , 2024, pp. 331–344

    2024

  23. [30]

    Agcv-loam: Air-ground cross-view based lidar odometry and mapping,

    M. Zhu, Y . Yang, W. Song, M. Wang, and M. Fu, “Agcv-loam: Air-ground cross-view based lidar odometry and mapping,” in 2020 Chinese Control And Decision Conference (CCDC) , 2020, pp. 5261– 5266

    2020

  24. [31]

    Cross-view geo-localization: A survey,

    A. Durgam, S. Paheding, V . Dhiman, and V . Devabhaktuni, “Cross-view geo-localization: A survey,” IEEE Access , vol. 12, p. 192028–192050, 2024

    2024

  25. [32]

    Autonomous vehicle localization without prior high-definition map,

    S. Lee and J.-H. Ryu, “Autonomous vehicle localization without prior high-definition map,” IEEE Transactions on Robotics , vol. 40, pp. 2888–2906, 2024

    2024

  26. [33]

    Road structure inspired ugv-satellite cross-view geo- localization,

    D. Hu, X. Yuan, H. Xi, J. Li, Z. Song, F. Xiong, K. Zhang, and C. Zhao, “Road structure inspired ugv-satellite cross-view geo- localization,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 17, pp. 16 767–16 786, 2024

    2024

  27. [34]

    Radar image-based positioning for usv under gps denial environment,

    H. Ma, E. Smart, A. Ahmed, and D. Brown, “Radar image-based positioning for usv under gps denial environment,” IEEE Transactions on Intelligent Transportation Systems , vol. 19, no. 1, pp. 72–80, 2018

    2018

  28. [35]

    Geo-localization with transformer-based 2d-3d match network,

    L. Li, Y . Ma, K. Tang, X. Zhao, C. Chen, J. Huang, J. Mei, and Y . Liu, “Geo-localization with transformer-based 2d-3d match network,”IEEE Robotics and Automation Letters , vol. 8, no. 8, pp. 4855–4862, 2023

    2023

  29. [36]

    Lidar-based vehicle localization on the satellite image via a neural network,

    M. Fu, M. Zhu, Y . Yang, W. Song, and M. Wang, “Lidar-based vehicle localization on the satellite image via a neural network,” Robotics and Autonomous Systems , vol. 129, p. 103519, 2020

    2020

  30. [38]

    A survey on sar and optical satellite image registration,

    O. Sommervold, M. Gazzea, and R. Arghandeh, “A survey on sar and optical satellite image registration,” Remote Sensing, vol. 15, no. 3, p. 850, 2023

    2023

  31. [39]

    Get it for free: Radar segmentation without expert labels and its application in odometry and localization,

    S. Li, Z. Hong, Y . Chen, L. Hu, and J. Qin, “Get it for free: Radar segmentation without expert labels and its application in odometry and localization,” IEEE Robotics and Automation Letters , vol. 10, no. 3, pp. 2678–2685, 2025

    2025

  32. [40]

    The finer points: A systematic comparison of point-cloud extractors for radar odometry,

    E. Preston-Krebs, D. Lisus, and T. D. Barfoot, “The finer points: A systematic comparison of point-cloud extractors for radar odometry,”

  33. [41]

    Attention u-net: Learning where to look for the pancreas,

    O. Oktay, J. Schlemper, L. L. Folgoc, M. Lee, M. Heinrich, K. Misawa, K. Mori, S. McDonagh, N. Y . Hammerla, B. Kainz, B. Glocker, and D. Rueckert, “Attention u-net: Learning where to look for the pancreas,” 2018. [Online]. Available: https://arxiv.org/abs/1804.03999

    2018 arXiv

  34. [42]

    Deep closest point: Learning represen- tations for point cloud registration,

    Y . Wang and J. M. Solomon, “Deep closest point: Learning represen- tations for point cloud registration,” 2019

    2019

  35. [43]

    borglab/gtsam,

    F. Dellaert and G. Contributors, “borglab/gtsam,” May 2022. [Online]. Available: https://github.com/borglab/gtsam)

    2022

  36. [44]

    Lidar-level localization with radar? the cfear ap- proach to accurate, fast, and robust large-scale radar odometry in diverse environments,

    D. Adolfsson, M. Magnusson, A. Alhashimi, A. J. Lilienthal, and H. Andreasson, “Lidar-level localization with radar? the cfear ap- proach to accurate, fast, and robust large-scale radar odometry in diverse environments,” IEEE Transactions on Robotics , vol. 39, no. 2, p. 1476–...

    2023

  37. [45]

    The oxford radar robotcar dataset: A radar extension to the oxford robotcar dataset,

    D. Barnes, M. Gadd, P. Murcutt, P. Newman, and I. Posner, “The oxford radar robotcar dataset: A radar extension to the oxford robotcar dataset,” in Proc. IEEE Int. Conf. Robot. Autom. (ICRA) , Paris, 2020

    2020

  38. [46]

    Adam: A method for stochastic optimiza- tion,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimiza- tion,” 2017

    2017

  39. [47]

    Boreas: A multi-season autonomous driving dataset,

    K. Burnett, D. J. Yoon, Y . Wu, A. Z. Li, H. Zhang, S. Lu, J. Qian, W.-K. Tseng, A. Lambert, K. Y . K. Leung, A. P. Schoellig, and T. D. Barfoot, “Boreas: A multi-season autonomous driving dataset,” 2023

    2023

  40. [48]

    A new method for the assessment of spatial accuracy and completeness of openstreetmap building foot- prints,

    M. A. Brovelli and G. Zamboni, “A new method for the assessment of spatial accuracy and completeness of openstreetmap building foot- prints,” ISPRS International Journal of Geo-Information , vol. 7, no. 8, 2018

    2018

  41. [50]

    Google, Google Maps JavaScript API v3 Reference , Google,

  42. [2020]

    Available: https://arxiv.org/abs/2001.10789

    [Online]. Available: https://arxiv.org/abs/2001.10789

    2001 arXiv

  43. [2024]

    Available: https://arxiv.org/abs/2409.12256

    [Online]. Available: https://arxiv.org/abs/2409.12256

    arXiv

  44. [2025]

    Available: https://developers.google.com/maps/ documentation/javascript/reference

    [Online]. Available: https://developers.google.com/maps/ documentation/javascript/reference

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.