pith. machine review for the scientific record. sign in

arxiv: 2604.24674 · v1 · submitted 2026-04-27 · 💻 cs.RO

Recognition: unknown

Pushing Radar Odometry Beyond the Pavement: Current Capabilities and Challenges

Shaunak Kolhe , Peng Jiang , Maggie Wigness , Philip Osteen , Timothy Overbye , Chrisitan Ellis , Srikanth Saripalli
Authors on Pith no claims yet

Pith reviewed 2026-05-08 02:52 UTC · model grok-4.3

classification 💻 cs.RO
keywords radar odometryoff-road roboticsmotion compensationIMU preintegrationSE(3) motiontrajectory estimationunstructured environments
0
0 comments X

The pith

Two radar baselines improve off-road trajectory estimates by handling full 3D motion and ground returns

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

This paper examines radar odometry outside urban pavement, where sensors must cope with complete three-dimensional vehicle motion, terrain reflections that clutter scans, and radar features that appear or disappear unpredictably. It introduces two minimal adaptations: one that compensates for vehicle motion while building three-dimensional radar point clouds, and another that folds in inertial preintegration to keep scan matching steady. Tests on a new outdoor dataset show measurable gains in pose accuracy along difficult routes. The work supplies concrete starting points rather than claiming a finished solution.

Core claim

By applying motion compensation to radar scans and preintegrating IMU measurements, standard scan-matching pipelines can produce usable trajectory estimates in off-road settings despite non-planar motion, dense ground returns, and unstable features, as verified on the Great Outdoors dataset.

What carries the argument

Radar-KISSICP baseline that generates 3D-aware point clouds via motion compensation, paired with Radar-IMU baseline that stabilizes matching through IMU preintegration

If this is right

  • Trajectory accuracy improves on routes with large elevation changes and rough surfaces
  • Radar becomes a practical sensor choice for localization when cameras and lidars lose visibility
  • The baselines establish a reference level against which more advanced off-road radar methods can be measured

Where Pith is reading between the lines

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

  • Future systems could fuse the compensated radar clouds with other modalities to further reduce drift in extended operations
  • The same motion-compensation step may generalize to other rotating sensors that must operate on sloped or bumpy ground
  • The dataset itself supplies a testbed for studying how feature density changes with terrain type and vehicle attitude

Load-bearing premise

The two proposed baselines are enough to overcome full SE(3) motion, terrain-induced ground returns, and sparse or unstable radar features.

What would settle it

Running the baselines on the Great Outdoors dataset routes that contain pronounced pitch, roll, and ground clutter yields no reduction in trajectory error compared with uncompensated radar matching.

Figures

Figures reproduced from arXiv: 2604.24674 by Chrisitan Ellis, Maggie Wigness, Peng Jiang, Philip Osteen, Shaunak Kolhe, Srikanth Saripalli, Timothy Overbye.

Figure 1
Figure 1. Figure 1: Motivating scenario for the use of radar in an off￾road, degraded environment. A Clearpath Warthog equipped with a Navtech radar operates in a snow covered dense forest. confounding rather than outliers, false correspondences accu￾mulate in ICP pipelines, and sparse radar pointclouds lead to ill-conditioned registration. To address these challenges, we evaluate existing radar odometry engines in off-road s… view at source ↗
Figure 2
Figure 2. Figure 2: Overhead view of routes from the GO dataset [8] (left) view at source ↗
Figure 3
Figure 3. Figure 3: Overlap of relative transforms in SE(3) on SE(2). While neither automotive nor offroad sequences exhibit motion restricted to a plane, the deviation from this assumption is greater in the offroad sequences as can be seen from the standard deviation of the data from unity. Table I summarizes the radar sensor specifications across datasets. To highlight the difference in motion regimes, we compute the overla… view at source ↗
Figure 4
Figure 4. Figure 4: Estimated trajectories from current RADAR odometry view at source ↗
Figure 5
Figure 5. Figure 5: Trajectory plots with overlayed Relative Pose Error view at source ↗
Figure 7
Figure 7. Figure 7: Isolated trajectory from Route 3 of the GO dataset. view at source ↗
Figure 8
Figure 8. Figure 8: Trajectory plots from Radar-KISSICP, CFEAR, view at source ↗
Figure 6
Figure 6. Figure 6: Heavy pitching from portion of Route 0 in the GO view at source ↗
Figure 9
Figure 9. Figure 9: Camera images and radar pointclouds from Route 3. Left: The Warthog is on level ground at the edge of the ravine view at source ↗
read the original abstract

Radar offers unique advantages for localization in unstructured environments, including robustness to weather, lighting, and airborne particulates. While most prior work has studied radar odometry in urban, largely planar settings, its performance in off-road environments remains less understood. In this paper, we investigate the potential of radar for off-road odometry estimation and identify key challenges that arise from full $SE(3)$ vehicle motion, terrain-induced ground returns, and sparse or unstable features. To address these issues, we introduce two simple baselines: Radar-KISSICP, which applies motion compensation to generate 3D-aware radar pointclouds, and Radar-IMU, which leverages IMU preintegration to stabilize scan matching. Experiments on the Great Outdoors (GO) dataset demonstrate that these baselines improve trajectory estimation in challenging routes and provide a reference point for future development of radar odometry in off-road robotics.

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.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract introduces no free parameters, axioms, or invented entities; it describes empirical baselines and qualitative challenges without derivations.

pith-pipeline@v0.9.0 · 5465 in / 1036 out tokens · 63489 ms · 2026-05-08T02:52:42.199745+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

27 extracted references · 15 canonical work pages

  1. [1]

    A New Wave in Robotics: Survey on Recent MmWave Radar Applications in Robotics,

    K. Harlow, H. Jang, T. D. Barfoot, A. Kim, and C. Heckman, “A New Wave in Robotics: Survey on Recent MmWave Radar Applications in Robotics,”IEEE Transactions on Robotics, vol. 40, pp. 4544–4560, 2024. [Online]. Available: https://ieeexplore.ieee.org/document/10683889

    work page arXiv 2024

  2. [2]

    Lidar-Level Localization With Radar? The CFEAR Approach 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 Approach to Accurate, Fast, and Robust Large-Scale Radar Odometry in Diverse Environments,”IEEE Transactions on Robotics, vol. 39, no. 2, pp. 1476–1495, Apr. 2023. [Online]. Available: https://ieeexplore.ieee.org/document/9969174

    work page arXiv 2023

  3. [3]

    ORORA: Outlier-Robust Radar Odometry,

    H. Lim, K. Han, G. Shin, G. Kim, S. Hong, and H. Myung, “ORORA: Outlier-Robust Radar Odometry,” in2023 IEEE International Conference on Robotics and Automation (ICRA), May 2023, pp. 2046–2053. [Online]. Available: https://ieeexplore.ieee.org/document/10160997

    work page arXiv 2023

  4. [4]

    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,” in2020 IEEE International Conference on Robotics and Automation (ICRA), May 2020, pp. 6433–6438, iSSN: 2577-087X. [Online]. Available: https://ieeexplore.ieee.org/document/9196884

    work page arXiv 2020

  5. [5]

    MulRan: Multimodal Range Dataset for Urban Place Recognition,

    G. Kim, Y . S. Park, Y . Cho, J. Jeong, and A. Kim, “MulRan: Multimodal Range Dataset for Urban Place Recognition,” in2020 IEEE International Conference on Robotics and Automation (ICRA), May 2020, pp. 6246–6253, iSSN: 2577-087X. [Online]. Available: https://ieeexplore.ieee.org/document/9197298

    work page arXiv 2020

  6. [6]

    Boreas: A Multi-Season Autonomous Driving Dataset , url =

    K. Burnett, D. J. Yoon, Y . Wu, A. Z. Li, H. Zhang, S. Lu, J. Qian, W.-K. Tseng, A. Lambert, K. Y . Leung, A. P. Schoellig, and T. D. Barfoot, “Boreas: A multi-season autonomous driving dataset,”The International Journal of Robotics Research, vol. 42, no. 1-2, pp. 33–42, Jan. 2023. [Online]. Available: https://journals.sagepub.com/doi/10.1177/02783649231160195

    work page doi:10.1177/02783649231160195 2023

  7. [7]

    Kiss-icp: In defense of point-to-point icp – simple, accurate, and robust registration if done the right way,

    I. Vizzo, T. Guadagnino, B. Mersch, L. Wiesmann, J. Behley, and C. Stachniss, “Kiss-icp: In defense of point-to-point icp – simple, accurate, and robust registration if done the right way,”IEEE Robotics and Automation Letters, vol. 8, no. 2, pp. 1029–1036, 2023

    2023

  8. [8]

    Go: The great outdoors multimodal dataset,

    P. Jiang, K. Viswanath, A. Nagariya, G. Chustz, M. Wigness, P. Osteen, T. Overbye, C. Ellis, L. Quang, and S. Saripalli, “Go: The great outdoors multimodal dataset,” 2025. [Online]. Available: https://arxiv.org/abs/2501.19274

    work page arXiv 2025

  9. [9]

    A Normal Distribution Transform-Based Radar Odometry Designed For Scanning and Automotive Radars,

    P.-C. Kung, C.-C. Wang, and W.-C. Lin, “A Normal Distribution Transform-Based Radar Odometry Designed For Scanning and Automotive Radars,” in2021 IEEE International Conference on Robotics and Automation (ICRA), May 2021, pp. 14 417–14 423, iSSN: 2577-087X. [Online]. Available: https://ieeexplore.ieee.org/document/9561413

    work page arXiv 2021

  10. [10]

    Probabilistic ego-motion estimation using multiple automotive radar sensors,

    M. Rapp, M. Barjenbruch, M. Hahn, J. Dickmann, and K. Dietmayer, “Probabilistic ego-motion estimation using multiple automotive radar sensors,”Robotics and Autonomous Systems, vol. 89, pp. 136–146, 2017. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0921889016300525

    2017

  11. [11]

    Real-time pose graph slam based on radar,

    M. Holder, S. Hellwig, and H. Winner, “Real-time pose graph slam based on radar,” in2019 IEEE Intelligent Vehicles Symposium (IV), 2019, pp. 1145–1151

    2019

  12. [12]

    Instantaneous ego-motion estimation using doppler radar,

    D. Kellner, M. Barjenbruch, J. Klappstein, J. Dickmann, and K. Di- etmayer, “Instantaneous ego-motion estimation using doppler radar,” in16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013), 2013, pp. 869–874

    2013

  13. [13]

    DRO: Doppler-Aware Direct Radar Odometry

    C. L. Gentil, L. Brizi, D. Lisus, X. Qiao, G. Grisetti, and T. D. Barfoot, “DRO: Doppler-Aware Direct Radar Odometry.”

  14. [14]

    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,”CoRR, vol. abs/2001.10789, 2020. [Online]. Available: https://arxiv.org/abs/2001.10789

    work page arXiv 2001

  15. [15]

    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,” inRobotics: Science and Systems, 2021

    2021

  16. [16]

    Pointing the way: Refining radar-lidar localization using learned icp weights,

    D. Lisus, J. Laconte, K. Burnett, Z. Zhang, and T. D. Barfoot, “Pointing the way: Refining radar-lidar localization using learned icp weights,” 2025. [Online]. Available: https://arxiv.org/abs/2309.08731

    work page arXiv 2025

  17. [17]

    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, vol. 10, no. 1, pp. 224–231, Jan. 2025. [Online]. Available: https://ieeexplore.ieee.org/document/10766427

    work page arXiv 2025

  18. [18]

    Ceres Solver,

    S. Agarwal, K. Mierle, and T. C. S. Team, “Ceres Solver,” 10 2023. [Online]. Available: https://github.com/ceres-solver/ceres-solver

    2023

  19. [19]

    Precise Ego-Motion Estimation with Millimeter-Wave Radar Under Diverse and Challenging Conditions,

    S. H. Cen and P. Newman, “Precise Ego-Motion Estimation with Millimeter-Wave Radar Under Diverse and Challenging Conditions,” in2018 IEEE International Conference on Robotics and Automation (ICRA), May 2018, pp. 6045–6052, iSSN: 2577-087X. [Online]. Available: https://ieeexplore.ieee.org/document/8460687

    work page arXiv 2018

  20. [20]

    ORB: An efficient alternative to SIFT or SURF,

    E. Rublee, V . Rabaud, K. Konolige, and G. Bradski, “ORB: An efficient alternative to SIFT or SURF,” in 2011 International Conference on Computer Vision, Nov. 2011, pp. 2564–2571, iSSN: 2380-7504. [Online]. Available: https://ieeexplore.ieee.org/document/6126544

    work page arXiv 2011

  21. [21]

    Object recognition from local scale-invariant features,

    D. Lowe, “Object recognition from local scale-invariant features,” inProceedings of the Seventh IEEE International Conference on Computer Vision, vol. 2, Sept. 1999, pp. 1150–1157 vol.2. [Online]. Available: https://ieeexplore.ieee.org/document/790410

    1999

  22. [22]

    Parallel Maximum Clique Algorithms with Applications to Network Analysis and Storage,

    R. A. Rossi, D. F. Gleich, A. H. Gebremedhin, and M. M. A. Patwary, “Parallel Maximum Clique Algorithms with Applications to Network Analysis and Storage,” Dec. 2013, arXiv:1302.6256 [cs]. [Online]. Available: http://arxiv.org/abs/1302.6256

    work page arXiv 2013

  23. [23]

    Forster, L

    C. Forster, L. Carlone, F. Dellaert, and D. Scaramuzza,IMU preinte- gration on Manifold for Efficient Visual-Inertial Maximum-a-Posteriori Estimation, 2015

    2015

  24. [24]

    borglab/gtsam,

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

    2022

  25. [25]

    Warthog Unmanned Ground Vehicle Robot - Clearpath

    “Warthog Unmanned Ground Vehicle Robot - Clearpath.” [Online]. Available: https://clearpathrobotics.com/warthog-unmanned- ground-vehicle-robot/

  26. [26]

    The International Journal of Robotics Research32(11), 1231–1237 (2013).https://doi.org/10.1177/0278364913491297,https: //journals.sagepub.com/doi/10.1177/02783649134912972

    A. Geiger, P. Lenz, C. Stiller, and R. Urtasun, “Vision meets robotics: The KITTI dataset,”The International Journal of Robotics Research, vol. 32, no. 11, pp. 1231–1237, Sept. 2013. [Online]. Available: https://journals.sagepub.com/doi/10.1177/0278364913491297

    work page doi:10.1177/0278364913491297 2013

  27. [27]

    evo: Python package for the evaluation of odometry and slam

    M. Grupp, “evo: Python package for the evaluation of odometry and slam.” https://github.com/MichaelGrupp/evo, 2017

    2017