pith. sign in

arxiv: 2604.13571 · v2 · submitted 2026-04-15 · 💻 cs.CV

Radar-Informed 3D Multi-Object Tracking under Adverse Conditions

Bingxue Xu , Emil Hedemalm , Ajinkya Khoche , Patric Jensfelt This is my paper

Pith reviewed 2026-05-10 13:41 UTC · model grok-4.3

classification 💻 cs.CV
keywords 3D multi-object trackingradar fusionadverse weatherlong-range trackingsensor fusionstate estimationpoint cloudsautonomous driving
0
0 comments X

The pith

Radar point clouds used as explicit observations improve 3D multi-object tracking by refining states and recovering missed objects at long range.

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

The paper introduces RadarMOT, a framework that feeds radar point clouds directly into the tracking process as additional observations to correct object states and recover detections the primary detector missed. This differs from typical multimodal fusion that embeds radar inside a learned network, where overall model degradation can erase radar's potential benefits. The goal is to sustain tracking performance when range increases or weather impairs cameras and LiDAR. Evaluations on a truck scene dataset show these explicit updates deliver clear accuracy gains under the targeted conditions.

Core claim

RadarMOT explicitly uses radar point clouds as additional observations to refine state estimation and recover objects missed by the detector, yielding a 12.7% improvement in Average Multi-Object Tracking Accuracy at long range and up to 10.3% in adverse weather on the MAN-TruckScenes dataset.

What carries the argument

The radar observation update step that integrates radar point clouds directly into the tracker to refine states and recover missed detections.

If this is right

  • Tracking accuracy improves by 12.7% at long ranges where detectors commonly fail.
  • Performance rises by up to 10.3% under adverse weather that degrades camera and LiDAR data.
  • Explicit radar use preserves robustness even when the overall fusion model would otherwise degrade.
  • The approach provides a modular way to add radar observations to existing 3D trackers.

Where Pith is reading between the lines

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

  • Explicit sensor integration may prove easier to debug and adapt than end-to-end learned fusion in safety-critical settings.
  • The same principle could be tested with other long-range modalities such as additional radar channels or thermal data.
  • Autonomous systems might gain reliability by routing radar observations through a separate non-learned path during low-visibility events.

Load-bearing premise

Radar point clouds can be treated as reliable, low-noise additional observations that refine state estimates and recover missed objects without introducing false positives or degrading performance when the primary detector is already accurate.

What would settle it

Measure AMOTA on the same dataset after removing or corrupting all radar points and confirm whether accuracy falls below the non-radar baseline tracker.

Figures

Figures reproduced from arXiv: 2604.13571 by Ajinkya Khoche, Bingxue Xu, Emil Hedemalm, Patric Jensfelt.

Figure 1
Figure 1. Figure 1: Overall framework of the proposed RadarMOT, built [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Comparison of radar point clouds before and after mo [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Radar-informed Kalman update. Purple points and ar [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison across range and adverse conditions on [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative comparison on TruckScenes in fog and at long distances. Green bounding boxes indicate the ground truth, [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
read the original abstract

The challenge of 3D multi-object tracking is achieving robustness in real-world applications, for example under adverse conditions and maintaining consistency as distance increases. To overcome these challenges, sensor fusion approaches that combine LiDAR, cameras, and radar have emerged. However, existing multimodal methods usually treat radar as another learned feature inside the network. When the overall model degrades in difficult environments, the robustness advantages that radar could provide are also reduced. In this paper we propose RadarMOT, a radar-informed 3D multi-object tracking framework that explicitly uses radar point clouds as additional observations to refine state estimation and recover objects missed by the detector at long ranges. Evaluations on the MAN-TruckScenes dataset show that RadarMOT consistently improves the Average Multi-Object Tracking Accuracy (AMOTA) by 12.7\% at long range and up to 10.3\% in adverse weather. The code will be available at https://github.com/bingxue-xu/radarmot

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 / 1 minor

Summary. The paper proposes RadarMOT, a radar-informed 3D multi-object tracking framework that explicitly uses radar point clouds as additional observations to refine state estimation and recover objects missed by the primary LiDAR/camera detector, particularly at long ranges and under adverse conditions. Evaluations on the MAN-TruckScenes dataset report AMOTA improvements of 12.7% at long range and up to 10.3% in adverse weather, with code to be released.

Significance. If the empirical claims hold after proper validation, the work has moderate significance for multimodal tracking in autonomous driving, as the explicit (rather than learned-feature) use of radar could preserve physical advantages when other sensors degrade. Code availability is a positive for reproducibility.

major comments (3)
  1. Abstract: the reported AMOTA gains of 12.7% at long range and 10.3% in adverse weather are presented without any description of the baseline trackers, ablation studies isolating radar's contribution, number of runs, or error bars, leaving the central quantitative claim only moderately supported.
  2. Method section: the description of how radar point clouds are associated, filtered, and used to update states or recover missed detections lacks concrete details on noise modeling, association thresholds, or clutter rejection, which is load-bearing for the assumption that radar acts as reliable low-noise observations.
  3. Experiments section: no ablation or analysis is provided for cases where the primary detector already has high recall; this is required to confirm that radar updates yield net gains rather than introducing false positives or state drift from increased clutter and multipath in adverse weather.
minor comments (1)
  1. Abstract: the GitHub link is given but no license or exact commit is specified, which is a minor reproducibility detail.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major point below and will revise the paper to incorporate additional details and analyses where appropriate to strengthen the presentation of our results.

read point-by-point responses

  1. Referee: Abstract: the reported AMOTA gains of 12.7% at long range and 10.3% in adverse weather are presented without any description of the baseline trackers, ablation studies isolating radar's contribution, number of runs, or error bars, leaving the central quantitative claim only moderately supported.

    Authors: We agree that the abstract provides limited context for the quantitative claims due to its brevity. We will revise the abstract to briefly identify the baseline as a standard LiDAR-based 3D MOT method and explicitly reference the Experiments section for ablation studies, evaluation protocol, and statistical details. We will also add error bars and clarify the number of evaluation runs in the main text and figures to better support the reported gains. revision: partial

  2. Referee: Method section: the description of how radar point clouds are associated, filtered, and used to update states or recover missed detections lacks concrete details on noise modeling, association thresholds, or clutter rejection, which is load-bearing for the assumption that radar acts as reliable low-noise observations.

    Authors: We acknowledge that more implementation specifics are needed. In the revised Method section, we will add concrete details including the radar point association criteria (e.g., Euclidean distance and radial velocity thresholds), filtering steps, noise modeling (e.g., assumed Gaussian distributions with variances derived from sensor specifications), association thresholds, and clutter rejection mechanisms (e.g., RCS-based filtering and spatial consistency checks). This will substantiate the reliability of radar as low-noise observations. revision: yes

  3. Referee: Experiments section: no ablation or analysis is provided for cases where the primary detector already has high recall; this is required to confirm that radar updates yield net gains rather than introducing false positives or state drift from increased clutter and multipath in adverse weather.

    Authors: We agree this analysis is valuable to demonstrate net benefits. We will add a dedicated ablation study or subsection in the Experiments section evaluating RadarMOT in high-recall detector scenarios. This will include metrics on false positive rates, state estimation drift, and AMOTA changes, with specific focus on adverse weather conditions to confirm that radar updates provide gains without introducing clutter-related issues. revision: yes

Circularity Check

0 steps flagged

No circularity in empirical method and evaluation

full rationale

The paper proposes RadarMOT as a tracking framework that incorporates radar point clouds as additional observations for state refinement and missed-object recovery, then reports empirical AMOTA gains on the MAN-TruckScenes dataset. No derivation chain, mathematical predictions, or first-principles results are claimed; the central results are direct experimental outcomes from dataset evaluation rather than quantities that reduce to fitted parameters or self-referential definitions. No load-bearing self-citations, ansatzes, or uniqueness theorems appear in the provided text that would create circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract provides no explicit parameters or invented entities; the approach implicitly rests on standard multi-object tracking assumptions about sensor observation models.

axioms (1)
  • domain assumption Radar returns supply accurate range, velocity, and position measurements that remain usable when camera and LiDAR data degrade.
    Central to the claim that radar observations can refine state estimates and recover missed objects under adverse conditions.

pith-pipeline@v0.9.0 · 5473 in / 1193 out tokens · 33314 ms · 2026-05-10T13:41:18.888340+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

47 extracted references · 47 canonical work pages

  1. [1]

    Center- based 3d object detection and tracking

    T. Yin, X. Zhou, and P. Kr ¨ahenb¨uhl, “Center-based 3d object detection and tracking.” [Online]. Available: http://arxiv.org/abs/2006.11275

    work page arXiv 2006

  2. [2]

    Far3D: Expanding the Horizon for Surround-view 3D Object Detection,

    X. Jiang, S. Li, Y . Liu, S. Wang, F. Jia, T. Wang, L. Han, and X. Zhang, “Far3D: Expanding the Horizon for Surround-view 3D Object Detection,” Dec. 2023, arXiv:2308.09616 [cs]. [Online]. Available: http://arxiv.org/abs/2308.09616

    work page arXiv 2023

  3. [3]

    Petr: Position embedding transformation for multi-view 3d object detection

    Y . Liu, T. Wang, X. Zhang, and J. Sun, “PETR: Position embedding transformation for multi-view 3d object detection.” [Online]. Available: http://arxiv.org/abs/2203.05625

    work page arXiv

  4. [4]

    Exploring object-centric temporal modeling for efficient multi-view 3d object detection

    S. Wang, Y . Liu, T. Wang, Y . Li, and X. Zhang, “Exploring object-centric temporal modeling for efficient multi-view 3d object detection.” [Online]. Available: http://arxiv.org/abs/2303.11926

    work page arXiv

  5. [5]

    arXiv preprint arXiv:2110.06922 (2021)

    Y . Wang, V . Guizilini, T. Zhang, Y . Wang, H. Zhao, and J. Solomon, “DETR3d: 3d object detection from multi-view images via 3d-to-2d queries.” [Online]. Available: http://arxiv.org/abs/2110.06922

    work page arXiv

  6. [6]

    L4dr: LiDAR-4dradar fusion for weather-robust 3d object detection

    X. Huang, Z. Xu, H. Wu, J. Wang, Q. Xia, Y . Xia, J. Li, K. Gao, C. Wen, and C. Wang, “L4dr: LiDAR-4dradar fusion for weather-robust 3d object detection.” [Online]. Available: http://arxiv.org/abs/2408.03677

    work page arXiv

  7. [7]

    CenterFusion: Center-based radar and camera fusion for 3d object detection,

    R. Nabati and H. Qi, “CenterFusion: Center-based radar and camera fusion for 3d object detection,” in2021 IEEE Winter Conference on Applications of Computer Vision (WACV). IEEE, pp. 1526–1535. [Online]. Available: https://ieeexplore.ieee.org/document/9423268/

    work page arXiv

  8. [8]

    Unleashing HyDRa: Hybrid fusion, depth consistency and radar for unified 3d perception

    P. Wolters, J. Gilg, T. Teepe, F. Herzog, A. Laouichi, M. Hofmann, and G. Rigoll, “Unleashing HyDRa: Hybrid fusion, depth consistency and radar for unified 3d perception.” [Online]. Available: http: //arxiv.org/abs/2403.07746

    work page arXiv

  9. [9]

    Hvdetfusion: A simple and robust camera-radar fusion framework.arXiv preprint arXiv:2307.11323, 2023

    K. Lei, Z. Chen, S. Jia, and X. Zhang, “HVDetFusion: A simple and robust camera-radar fusion framework.” [Online]. Available: http://arxiv.org/abs/2307.11323

    work page arXiv

  10. [10]

    MixedFusion: An efficient multimodal data fusion framework for 3-d object detection and tracking,

    C. Zhang, H. Wang, L. Chen, Y . Li, and Y . Cai, “MixedFusion: An efficient multimodal data fusion framework for 3-d object detection and tracking,” vol. 36, no. 1, pp. 1842–1856. [Online]. Available: https://ieeexplore.ieee.org/document/10304295/

    work page arXiv

  11. [11]

    CR3DT: Camera-RADAR Fusion for 3D Detection and Tracking,

    N. Baumann, M. Baumgartner, E. Ghignone, J. K ¨uhne, T. Fischer, Y .-H. Yang, M. Pollefeys, and M. Magno, “CR3DT: Camera-RADAR Fusion for 3D Detection and Tracking,” in2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Oct. 2024, pp. 4926–4933. [Online]. Available: https://ieeexplore.ieee.org/abstract/ document/10801848

    work page arXiv 2024

  12. [12]

    FARFusion v2: A geometry-based radar-camera fusion method on the ground for roadside far-range 3d object detection

    Y . Li, J. Deng, Y . Xiao, Y . Wang, X. Chu, J. Ji, and Y . Zhang, “FARFusion v2: A geometry-based radar-camera fusion method on the ground for roadside far-range 3d object detection.”

    work page

  13. [13]

    Sparc: Sparse radar-camera fusion for 3d object detection.arXiv preprint arXiv:2411.19860, 2024

    P. Wolters, J. Gilg, T. Teepe, F. Herzog, F. Fent, and G. Rigoll, “SpaRC: Sparse radar-camera fusion for 3d object detection.” [Online]. Available: http://arxiv.org/abs/2411.19860

    work page arXiv

  14. [14]

    arXiv preprint arXiv:2008.08063 (2020)

    X. Weng, J. Wang, D. Held, and K. Kitani, “AB3dmot: A baseline for 3d multi-object tracking and new evaluation metrics.” [Online]. Available: http://arxiv.org/abs/2008.08063

    work page arXiv 2008

  15. [15]

    MCTrack: A unified 3d multi-object tracking framework for autonomous driving

    X. Wang, S. Qi, J. Zhao, H. Zhou, S. Zhang, G. Wang, K. Tu, S. Guo, J. Zhao, J. Li, and M. Yang, “MCTrack: A unified 3d multi-object tracking framework for autonomous driving.” [Online]. Available: http://arxiv.org/abs/2409.16149

    work page arXiv

  16. [16]

    Probabilistic 3d multi-object tracking for autonomous driving

    H.-k. Chiu, A. Prioletti, J. Li, and J. Bohg, “Probabilistic 3d multi-object tracking for autonomous driving.” [Online]. Available: http://arxiv.org/abs/2001.05673

    work page arXiv 2001

  17. [17]

    SimpleTrack: Understanding and rethinking 3d multi-object tracking

    Z. Pang, Z. Li, and N. Wang, “SimpleTrack: Understanding and rethinking 3d multi-object tracking.” [Online]. Available: http: //arxiv.org/abs/2111.09621

    work page arXiv

  18. [18]

    Easy-poly: An easy polyhedral framework for 3d multi-object tracking

    P. Zhang, X. Li, X. Lin, and L. He, “Easy-poly: An easy polyhedral framework for 3d multi-object tracking.” [Online]. Available: http://arxiv.org/abs/2502.17822

    work page arXiv

  19. [19]

    K. V . Ramachandra,Kalman Filtering Techniques for Radar Tracking. CRC Press

    work page

  20. [20]

    Naval Research Logistics Quarterly2(1–2), 83–97 (1955) https://doi.org/10.1002/nav

    H. W. Kuhn, “The hungarian method for the assignment problem,” Naval Research Logistics Quarterly, vol. 2, no. 1-2, pp. 83–97, 1955. [Online]. Available: https://onlinelibrary.wiley.com/doi/abs/10.1002/nav. 3800020109

    work page doi:10.1002/nav 1955

  21. [21]

    Distance-IoU loss: Faster and better learning for bounding box regression,

    Z. Zheng, P. Wang, W. Liu, J. Li, R. Ye, and D. Ren, “Distance-IoU loss: Faster and better learning for bounding box regression,” vol. 34, no. 7, pp. 12 993–13 000. [Online]. Available: https://ojs.aaai.org/index.php/AAAI/article/view/6999

    work page

  22. [22]

    nuScenes: A Multimodal Dataset for Autonomous Driving,

    H. Caesar, V . Bankiti, A. H. Lang, S. V ora, V . E. Liong, Q. Xu, A. Krishnan, Y . Pan, G. Baldan, and O. Beijbom, “nuScenes: A multimodal dataset for autonomous driving,” in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, pp. 11 618–11 628. [Online]. Available: https://ieeexplore.ieee.org/document/9156412/

    work page arXiv 2020

  23. [23]

    Scalability in perception for autonomous driving: Waymo open dataset,

    P. Sun, H. Kretzschmar, X. Dotiwalla, A. Chouard, V . Patnaik, P. Tsui, J. Guo, Y . Zhou, Y . Chai, B. Caine, V . Vasudevan, W. Han, J. Ngiam, H. Zhao, A. Timofeev, S. Ettinger, M. Krivokon, A. Gao, A. Joshi, Y . Zhang, J. Shlens, Z. Chen, and D. Anguelov, “Scalability in perception for autonomous driving: Waymo open dataset,” in 2020 IEEE/CVF Conference ...

    work page arXiv 2020

  24. [24]

    Poly-MOT: A polyhedral framework for 3d multi-object tracking,

    X. Li, T. Xie, D. Liu, J. Gao, K. Dai, Z. Jiang, L. Zhao, and K. Wang, “Poly-MOT: A polyhedral framework for 3d multi-object tracking,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 9391–9398. [Online]. Available: https://ieeexplore.ieee.org/document/10341778/

    work page arXiv

  25. [25]

    Fast-poly: A fast polyhedral framework for 3d multi-object tracking

    X. Li, D. Liu, Y . Wu, X. Wu, L. Zhao, and J. Gao, “Fast-poly: A fast polyhedral framework for 3d multi-object tracking.” [Online]. Available: http://arxiv.org/abs/2403.13443

    work page arXiv

  26. [26]

    IMM-MOT: A novel 3d multi-object tracking framework with interacting multiple model filter

    X. Liu, X. Zhao, G. Liu, Z. Wu, T. Wang, L. Meng, and Y . Wang, “IMM-MOT: A novel 3d multi-object tracking framework with interacting multiple model filter.” [Online]. Available: http: //arxiv.org/abs/2502.09672

    work page arXiv

  27. [27]

    Are we ready for autonomous driving? the KITTI vision benchmark suite,

    A. Geiger, P. Lenz, and R. Urtasun, “Are we ready for autonomous driving? the KITTI vision benchmark suite,” in2012 IEEE Conference on Computer Vision and Pattern Recognition, pp. 3354–3361. [Online]. Available: https://ieeexplore.ieee.org/document/6248074/

    work page arXiv

  28. [28]

    TrackFormer: Multi-object tracking with transformers,

    T. Meinhardt, A. Kirillov, L. Leal-Taixe, and C. Feichtenhofer, “TrackFormer: Multi-object tracking with transformers,” in2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, pp. 8834–8844. [Online]. Available: https://ieeexplore. ieee.org/document/9879668/

    work page arXiv

  29. [29]

    MUTR3d: A multi-camera tracking framework via 3d-to-2d queries,

    T. Zhang, X. Chen, Y . Wang, Y . Wang, and H. Zhao, “MUTR3d: A multi-camera tracking framework via 3d-to-2d queries,” in2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). IEEE, pp. 4536–4545. [Online]. Available: https://ieeexplore.ieee.org/document/9857405/

    work page arXiv

  30. [30]

    Standing between past and future: Spatio-temporal modeling for multi-camera 3d multi-object tracking,

    Z. Pang, J. Li, P. Tokmakov, D. Chen, S. Zagoruyko, and Y .-X. Wang, “Standing between past and future: Spatio-temporal modeling for multi-camera 3d multi-object tracking,” in2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, pp. 17 928–17 938. [Online]. Available: https: //ieeexplore.ieee.org/document/10204316/

    work page arXiv

  31. [31]

    CC-3dt: Panoramic 3d object tracking via cross-camera fusion

    T. Fischer, Y .-H. Yang, S. Kumar, M. Sun, and F. Yu, “CC-3dt: Panoramic 3d object tracking via cross-camera fusion.” [Online]. Available: http://arxiv.org/abs/2212.01247

    work page arXiv

  32. [32]

    DINO-MOT: 3d multi-object tracking with visual foundation model for pedestrian re-identification using visual memory mechanism,

    M. Y . Lee, C. D. W. Lee, J. Li, and M. H. Ang, “DINO-MOT: 3d multi-object tracking with visual foundation model for pedestrian re-identification using visual memory mechanism,” vol. 10, no. 2, pp. 1202–1208. [Online]. Available: https://ieeexplore.ieee.org/document/ 10755965/

    work page

  33. [33]

    MotionTrack: End-to-end transformer-based multi-object tracing with LiDAR-camera fusion

    C. Zhang, C. Zhang, Y . Guo, L. Chen, and M. Happold, “MotionTrack: End-to-end transformer-based multi-object tracing with LiDAR-camera fusion.” [Online]. Available: http://arxiv.org/abs/2306.17000

    work page arXiv

  34. [34]

    TransTrack: Multiple-object tracking with transformer,

    P. Sun, J. Cao, Y . Jiang, R. Zhang, E. Xie, Z. Yuan, C. Wang, and P. Luo, “TransTrack: Multiple object tracking with transformer.” [Online]. Available: http://arxiv.org/abs/2012.15460

    work page arXiv 2012

  35. [35]

    In: European Conference on Computer Vision, pp

    F. Zeng, B. Dong, Y . Zhang, T. Wang, X. Zhang, and Y . Wei, “MOTR: End-to-end multiple-object tracking with transformer.” [Online]. Available: http://arxiv.org/abs/2105.03247

    work page arXiv

  36. [36]

    Motrv3: Release-fetch supervision for end-to-end multi-object tracking

    E. Yu, T. Wang, Z. Li, Y . Zhang, X. Zhang, and W. Tao, “MOTRv3: Release-fetch supervision for end-to-end multi-object tracking.” [Online]. Available: http://arxiv.org/abs/2305.14298

    work page arXiv

  37. [37]

    MOTRv2: Bootstrapping end- to-end multi-object tracking by pretrained object detectors

    Y . Zhang, T. Wang, and X. Zhang, “MOTRv2: Bootstrapping end- to-end multi-object tracking by pretrained object detectors.” [Online]. Available: http://arxiv.org/abs/2211.09791

    work page arXiv

  38. [38]

    Exploring Radar Data Representations in Autonomous Driving: A Comprehensive Review,

    S. Yao, R. Guan, Z. Peng, C. Xu, Y . Shi, W. Ding, E. G. Lim, Y . Yue, H. Seo, K. L. Man, J. Ma, X. Zhu, and Y . Yue, “Exploring radar data representations in autonomous driving: A comprehensive review.” [Online]. Available: http://arxiv.org/abs/2312.04861

    work page arXiv

  39. [39]

    Radar and camera fusion for object detection and tracking: A comprehensive survey,

    K. Shi, S. He, Z. Shi, A. Chen, Z. Xiong, J. Chen, and J. Luo, “Radar and camera fusion for object detection and tracking: A comprehensive survey,”IEEE Communications Surveys & Tutorials, 2025

    work page 2025

  40. [40]

    CFTrack: Center-based radar and camera fusion for 3d multi-object tracking

    R. Nabati, L. Harris, and H. Qi, “CFTrack: Center-based radar and camera fusion for 3d multi-object tracking.” [Online]. Available: http://arxiv.org/abs/2107.05150

    work page arXiv

  41. [41]

    RaTrack: Moving object detection and tracking with 4d radar point cloud,

    Z. Pan, F. Ding, H. Zhong, and C. X. Lu, “RaTrack: Moving object detection and tracking with 4d radar point cloud,” in2024 IEEE International Conference on Robotics and Automation (ICRA), pp. 4480–4487. [Online]. Available: https://ieeexplore.ieee.org/document/ 10610368/

    work page

  42. [42]

    FARFusion: A practical roadside radar-camera fusion system for far-range perception,

    Y . Li, Y . Wang, C. Meng, Y . Duan, J. Ji, Y . Zhang, and Y . Zhang, “FARFusion: A practical roadside radar-camera fusion system for far-range perception,” vol. 9, no. 6, pp. 5433–5440. [Online]. Available: https://ieeexplore.ieee.org/document/10496834/

    work page arXiv

  43. [43]

    Simpler is better: Revisiting doppler velocity for enhanced moving object tracking with FMCW LiDAR,

    Y . Zeng, T. Wu, S. Qi, J. Li, X. Duan, and Y . Yu, “Simpler is better: Revisiting doppler velocity for enhanced moving object tracking with FMCW LiDAR,” in2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 7842–7849. [Online]. Available: https://ieeexplore.ieee.org/document/11246618/

    work page arXiv

  44. [44]

    MAN TruckScenes: A multimodal dataset for autonomous trucking in diverse conditions

    F. Fent, F. Kuttenreich, F. Ruch, F. Rizwin, S. Juergens, L. Lechermann, C. Nissler, A. Perl, U. V oll, M. Yan, and M. Lienkamp, “MAN TruckScenes: A multimodal dataset for autonomous trucking in diverse conditions.” [Online]. Available: http://arxiv.org/abs/2407.07462

    work page arXiv

  45. [45]

    Himo: High-speed objects motion compensation in point clouds,

    Q. Zhang, A. Khoche, Y . Yang, L. Ling, S. S. Mansouri, O. Andersson, and P. Jensfelt, “Himo: High-speed objects motion compensation in point clouds,”IEEE Transactions on Robotics, 2025

    work page 2025

  46. [46]

    Thrun, W

    S. Thrun, W. Burgard, and D. Fox,Probabilistic robotics. Cambridge, Mass.: MIT Press, 2005. [Online]. Available: http://www.amazon.de/gp/product/0262201623/102-8479661-9831324? v=glance&n=283155&n=507846&s=books&v=glance

    work page arXiv 2005

  47. [47]

    nuscenes tracking task,

    nuScenes Team, “nuscenes tracking task,” Online, 2020, accessed: 2026-04-09. [Online]. Available: https://www.nuscenes.org/tracking

    work page 2020