Pith. sign in

REVIEW 3 major objections 4 minor 41 references

A complete pipeline from raw mmWave radar chirps through synthetic-aperture focusing produces occupancy maps that outline indoor free space well enough for A* path planning and beat common radar baselines.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-14 13:48 UTC pith:7ZPRPUKK

load-bearing objection Solid end-to-end SAR-to-occupancy pipeline from raw cascaded mmWave IF, with open data; gains real under oracle poses but untested under realistic odometry noise. the 3 major comments →

arxiv 2607.10161 v1 pith:7ZPRPUKK submitted 2026-07-11 cs.RO

Millimeter Wave Radar: From Synthetic Aperture to Probabilistic Mapping

Jui-Te Huang , Ruoyang Xu , Michael Kaess This is my paper
classification cs.RO
keywords mmWave radarsynthetic aperture radarprobabilistic occupancy mappingFMCWback-projectionpath planningrobotics
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Robots that must navigate through smoke or fog cannot rely on LiDAR or cameras, so the paper asks whether millimeter-wave radar can still produce usable probabilistic occupancy maps. It answers by chaining three steps: coherent synthetic-aperture integration of every raw intermediate-frequency sample into a complex scattering image, conversion of that image’s amplitude into occupancy probability via a Rayleigh model, and ray-cast log-odds updates that respect occlusion. On ten indoor sequences the resulting maps recover corridor geometry more consistently than CFAR point clouds or simple range-azimuth heatmaps, and the A* paths planned on them succeed more often when checked against LiDAR ground truth. The work therefore shows that classical radar coherence, properly turned into probabilities, is already sufficient for downstream navigation without neural networks or manufacturer-specific training data.

Core claim

When every raw FMCW sample is phase-corrected and summed by the back-projection formula, the resulting complex map can be turned into a bounded occupancy probability by a Rayleigh CDF followed by transmittance-weighted log-odds updates; the maps so obtained have lower Chamfer distance and higher F-score than CFAR or range-azimuth baselines and support higher radar-to-LiDAR path-planning success rates across ten indoor corridors and open spaces.

What carries the argument

The back-projection sum that coherently integrates every intermediate-frequency sample into a complex grid M (Eq. 11), followed by the Rayleigh-normalized amplitude R and the transmittance-modulated log-odds update of the occupancy map L (Eqs. 12–17).

Load-bearing premise

Every transmit and receive antenna pose must be known accurately enough that the phase-correction term stays coherent; any unmodeled drift or timing error defocuses the synthetic-aperture image and collapses the probability model.

What would settle it

Re-run the identical ten sequences with deliberately injected pose noise of a few centimeters or a few degrees and measure whether Chamfer distance and A* success rates fall below the CFAR and range-azimuth baselines.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Indoor robots can build navigation-grade occupancy maps from commercial cascade mmWave radars without LiDAR supervision or learned denoisers.
  • Only three of the four cascade devices (108 virtual antenna pairs) are needed to reach the performance plateau, cutting computation and array size.
  • The same Rayleigh-plus-transmittance pipeline can be applied to any FMCW radar whose raw IF samples and poses are available, independent of manufacturer.
  • Downstream planners can treat the resulting free-space probability as a drop-in replacement for LiDAR occupancy maps in smoke-filled environments.

Where Pith is reading between the lines

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

  • Because the method never trains on LiDAR, it should transfer to outdoor or multi-floor scenes once accurate poses are supplied, offering a non-learning alternative to recent neural radar mappers.
  • Tightening the timing synchronization between chirps and odometry would directly raise the upper bound on map fidelity, suggesting a hardware rather than algorithmic next step.
  • The 30-degree delayed field-of-view used for occupancy updates is a pragmatic focus mask; replacing it with a learned or adaptive focus metric could recover more of the full aperture without introducing unfocused free-space holes.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper presents a complete pipeline that converts raw FMCW intermediate-frequency signals from cascaded mmWave radars into 2-D probabilistic occupancy maps. It first forms a complex-valued SAR image via back-projection (Eq. 11), converts amplitudes to occupancy probabilities with a Rayleigh CDF (Eq. 12), then performs ray-casting and log-odds updates (Eqs. 13–17) that respect transmittance and multi-path. The method is evaluated on ten indoor sequences against CFAR point-cloud and Range-Azimuth heatmap baselines using Chamfer/Hausdorff/F-score metrics (Table III) and A* path-planning success rates (Table IV, RoL/LoR). Ablations examine the Rayleigh scale σ_a and the number of antenna devices; an open-source cascaded-radar dataset and GPU pipeline are released.

Significance. If the reported gains hold under realistic conditions, the work supplies a practical, non-learning route from automotive-grade radar to occupancy maps usable for planning in smoke/fog, where LiDAR and cameras fail. The open dataset, GPU-accelerated code, systematic comparison of three signal-processing routes, and explicit downstream planning metric are concrete contributions that lower the barrier for subsequent radar-mapping research. The antenna-count ablation further gives useful engineering guidance on the diminishing returns of denser arrays.

major comments (3)
  1. [§III, Eq. (11); §IV-B1] §III and §IV-B1 state that all Tx/Rx antenna poses are known a priori and are obtained by interpolating FAST-LIO or openVINS trajectories. Every term of the coherent sum in Eq. 11 multiplies the measured sample by the phase factor exp(-j 4π f d / c). At 77 GHz a few millimetres of pose error rotates the phasor enough to destroy constructive addition at true scatterers and to inflate residual amplitudes that later feed the Rayleigh CDF (Eq. 12). No experiment injects centimetre-level drift typical of radar-only or radar-inertial odometry; consequently the quantitative superiority shown in Tables III–IV is demonstrated only under an oracle-localisation regime that the method itself does not provide. A controlled pose-noise study (or at least a discussion of the coherence length) is required before the claimed advantage over non-coherent baselines can be regarded as established for real rob
  2. [Abstract; §I; §IV-A] The abstract and introduction motivate the work by the failure of optical sensors in smoke and fog, yet every sequence (Fig. 5, Tables III–IV) is collected in clear indoor corridors and open workspaces. Without at least one controlled adverse-condition trial, it remains unproven that the SAR-plus-Rayleigh pipeline retains its reported edge precisely when the claimed robustness is most needed.
  3. [§III-C] The two-frame delay and the restriction of occupancy updates to a 30° forward FOV (§III-C) are introduced as ad-hoc safeguards against unfocused SAR regions. No sensitivity analysis is supplied; if these heuristics are removed or altered, the free-space versus unknown-space distinction (and therefore the planning success rates of Table IV) may change substantially. A short ablation or theoretical justification is needed to show that the reported gains are not artefacts of these choices.
minor comments (4)
  1. [Tables III–IV] Tables III and IV report point estimates without error bars or statistical tests across the ten sequences; a simple paired Wilcoxon or bootstrap interval would strengthen the claim of consistent superiority.
  2. [Fig. 4; Table II] Figure 4 shows the F-score peak at σ_a ≈ 0.15, yet the same value is used for all sequences without reporting per-sequence variation; a brief note on transferability would help readers re-calibrate for new sensors.
  3. [Eqs. (12)–(17)] The notation for the complex map M(q) and the normalised amplitude |M|/M_n is introduced cleanly, but the subsequent conversion to log-odds L(n) re-uses the symbol R for both the Rayleigh probability and the ray index; a distinct symbol would avoid momentary confusion.
  4. [§IV-E] In the path-planning protocol the maps are inflated by a fixed 20 cm; stating the robot’s actual footprint or showing a sensitivity plot would make the RoL/LoR numbers easier to interpret.

Circularity Check

1 steps flagged

No load-bearing circularity; empirical pipeline validated on independent LiDAR maps, with only minor hyperparameter selection of Rayleigh scale on the reported F-score curves.

specific steps
  1. fitted input called prediction [§IV-C Parameter Details / Fig. 4]
    "We perform a parameter search over σ_a and evaluate the F-score against the ground-truth occupancy map across multiple sequences, as shown in Figure 4. The best average performance is achieved at σ_a ≈0.15."

    σ_a is chosen by maximizing the identical F-score metric later tabulated for the proposed method on the same ten sequences. Absolute F-scores for the proposed pipeline are therefore mildly optimized rather than purely held-out; the effect is limited because the value is fixed once, shared with the RA baseline that uses the same Rayleigh model, and does not alter the qualitative ranking or planning results.

full rationale

The paper presents an engineering pipeline (FMCW IF signals to back-projection SAR via Eq. 11, Rayleigh CDF normalization Eq. 12, transmittance-weighted log-odds occupancy Eqs. 13–17) and evaluates it empirically against CFAR and Range-Azimuth baselines on ten indoor sequences using external LiDAR occupancy grids (Chamfer/Hausdorff/F-score in Table III) plus cross-validated A* planning success (RoL/LoR in Table IV). Poses are supplied by independent FAST-LIO/openVINS odometry; no equation equates the claimed map superiority to an input by construction. The sole minor issue is selection of the Rayleigh scale σ_a by grid search that maximizes average F-score on the same evaluation sequences (Fig. 4, §IV-C); the chosen value is then frozen and applied uniformly, and the same Rayleigh model is also used for the RA baseline, so the ranking versus baselines is not forced. No self-definitional loops, uniqueness theorems imported from the authors, ansatz smuggling, or renaming of known results appear. The work is self-contained against external benchmarks and does not claim first-principles predictions that reduce to fitted inputs.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The central empirical claim rests on standard radar physics, a classical occupancy-mapping recursion, a noise model whose single free scale is fitted once, and the strong operational assumption that antenna poses are known to high accuracy. No new physical entities are postulated; the free parameters are the usual occupancy clamps plus the Rayleigh scale.

free parameters (3)
  • σ_a (Rayleigh scale) = 0.15
    Controls the conversion of normalized SAR amplitude into occupancy probability (Eq. 12). Chosen by grid search that maximizes average F-score across the ten evaluation sequences (Fig. 4); value 0.15 adopted for all subsequent experiments.
  • p_hit / p_miss occupancy clamps = 0.7 / 0.2
    Upper and lower probability bounds used inside the log-odds update (Table II). Hand-chosen conventional values that prevent saturation.
  • l_max / l_min log-odds clamps = 3.5 / −2.0
    Hard bounds on the stored map values (Table II) that keep the map adaptable. Hand-chosen.
axioms (4)
  • domain assumption Antenna poses (and therefore two-way distances d) are known accurately enough for coherent phase correction in the SAR sum.
    Stated at the opening of §III and realized by interpolating FAST-LIO / openVINS poses; without it the complex map M does not focus.
  • domain assumption Normalized SAR amplitude follows a Rayleigh distribution whose scale equals the sensor thermal noise floor.
    Invoked in §III-C (Eq. 12) with citation to classical SAR literature; the scale itself is fitted.
  • standard math Standard log-odds occupancy recursion (Eqs. 14–16) correctly fuses successive independent sensor readings.
    Taken from the OctoMap / occupancy-grid literature [1] and applied after transmittance weighting.
  • ad hoc to paper A two-frame delay and a 30° forward FOV suffice to keep only focused SAR regions in the occupancy update.
    Engineering choice stated in the last paragraph of §III-C; not derived from first principles.

pith-pipeline@v1.1.0-grok45 · 17243 in / 2895 out tokens · 29440 ms · 2026-07-14T13:48:47.274503+00:00 · methodology

0 comments
read the original abstract

Robust probabilistic mapping is essential for autonomous robotic systems operating in challenging environments. While traditional sensors fail in adverse conditions such as smoke and fog, millimeter wave (mmWave) radar sensors offer reliable sensing in such conditions. However, creating accurate probabilistic maps from radar data presents significant challenges due to the inherently sparse and noisy characteristics of radio wave measurements and signal processing steps. In an attempt to address these issues, we establish a complete pipeline from raw radar signals to probabilistic occupancy maps, incorporating Synthetic Aperture Radar processing followed by a probabilistic modeling step. We conduct extensive validation across indoor environments, comparing our approach against different signal processing and probabilistic modeling approaches. We also evaluate mapping quality through downstream path planning performance analysis. Furthermore, we investigate the impact of key parameters and antenna array configuration on mapping performance. The experimental results demonstrate both the effectiveness and limitations of SAR-based probabilistic mapping for real-world robotic deployment. To facilitate future research and broader adoption, we contribute an open-source cascaded mmWave radar dataset with an accompanying GPU-accelerated signal processing pipeline available at https://github.com/rpl-cmu/rpm.

Figures

Figures reproduced from arXiv: 2607.10161 by Jui-Te Huang, Michael Kaess, Ruoyang Xu.

Figure 1
Figure 1. Figure 1: A demonstration of our proposed method using two cascade mmWave radar boards to create a LiDAR-like occupancy map (bottom right). As a vehicle moves through the environment, onboard radar sensors create a synthetic aperture to map the surroundings (Top). A probability modeling method is presented to create the occupancy map (bottom left). While research on radar-based state estimation is achiev￾ing perform… view at source ↗
Figure 2
Figure 2. Figure 2: Radar mapping procedures from raw signal to a probability map. In this work, we demonstrate using different signal processing steps to generate a probability map. We can jointly process multiple chirps of one radar frame or process each chirp separately to create a synthetic aperture for mapping. to the probability values on R. This step also helps us remove the artifacts created by the multi-path signal. … view at source ↗
Figure 3
Figure 3. Figure 3: The ground vehicle and sensors used for our experiments. Two cascade mmWave radars are installed on both sides of the vehicle. TABLE I: Signal Configuration for Cascade Radar Sensor Fstart Fslope # of RX # of TX Config 77 GHz 79 MHz/µs 16 12 chirp / Tx / frame samples/chirp chirp time Config 64 256 40 µs B. Ground Truth and Baselines 1) Ground Truth Pose: To test our mapping algorithm independently of pose… view at source ↗
Figure 5
Figure 5. Figure 5: Example probabilistic mapping outcomes of sequences used in experiments. Our proposed approach generated maps that clearly outlined the environment and distinguished between free space and unknown space. TABLE III: Obstacle Evaluation Results north north_short north_a north_b apart Method CD↓ HD↓ F-score↑ CD↓ HD↓ F-score↑ CD↓ HD↓ F-score↑ CD↓ HD↓ F-score↑ CD↓ HD↓ F-score↑ CFAR 1.6 8.605 0.382 2.92 9.265 0.… view at source ↗
Figure 6
Figure 6. Figure 6: Example path overlaid on their respective maps and LiDAR map. Proposed (top left, ), RA (top right, ), CFAR (bottom left, ), and LiDAR Map (bottom right). Grey: unknown; Black: navigable; White: obstacles. uniformly sample 200 pairs of valid start and end points in cells unoccupied across all maps, and use A* planner with Euclidean distance heuristics on 8-connected grid to generate paths on both radar and… view at source ↗
Figure 7
Figure 7. Figure 7: Mapping quality as a function of the number of antennas used. Each device contributes an additional 4 receivers and 3 transmitters. [10] J.-T. Huang, R. Xu, A. Hinduja, and M. Kaess, “Multi-radar inertial odometry for 3d state estimation using mmWave imaging radar,” in Proc. IEEE Intl. Conf. on Robotics and Automation (ICRA), Yoko￾hama, Japan, May 2024, pp. 12 006–12 012. [11] T. Huang, J. Miller, A. Prabh… view at source ↗

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

41 extracted references · 1 linked inside Pith

  1. [1]

    OctoMap: An efficient probabilistic 3D mapping framework based on octrees,

    A. Hornung, K. M. Wurm, M. Bennewitz, C. Stachniss, and W. Bur- gard, “OctoMap: An efficient probabilistic 3D mapping framework based on octrees,”Autonomous Robots, 2013, software available at https://octomap.github.io

    2013

  2. [2]

    RODNet: Radar object detection using cross-modal supervision,

    Y . Wang, Z. Jiang, X. Gao, J.-N. Hwang, G. Xing, and H. Liu, “RODNet: Radar object detection using cross-modal supervision,” in Proc. IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), Waikoloa, USA, Jan. 2021, pp. 504–513

    2021

  3. [3]

    ROD2021 challenge: A summary for radar object detection challenge for autonomous driving applications,

    Y . Wang, J.-N. Hwang, G. Wang, H. Liu, K.-J. Kim, H.-M. Hsu, J. Cai, H. Zhang, Z. Jiang, and R. Gu, “ROD2021 challenge: A summary for radar object detection challenge for autonomous driving applications,” inProc. ACM Intl. Conf. on Multimedia Retrieval (ICMR), Taipei, TW, Nov. 2021, pp. 553–559

    2021

  4. [4]

    Vision meets mmWave radar: 3d object perception benchmark for autonomous driving,

    Y . Wang, J.-H. Cheng, J.-T. Huang, S.-Y . Kuan, Q. Fu, C. Ni, S. Hao, G. Wang, G. Xing, H. Liu,et al., “Vision meets mmWave radar: 3d object perception benchmark for autonomous driving,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 2769–2775

    2024

  5. [5]

    Cross-modal contrastive learning of representations for navigation using lightweight, low-cost millimeter wave radar for adverse environmental conditions,

    J.-T. Huang, C.-L. Lu, P.-K. Chang, C.-I. Huang, C.-C. Hsu, P.- J. Huang, H.-C. Wang,et al., “Cross-modal contrastive learning of representations for navigation using lightweight, low-cost millimeter wave radar for adverse environmental conditions,”IEEE Robotics and Automation Letters (RA-L), vol. 6, no. 2, pp. 3333–3340, 2021

    2021

  6. [6]

    3D ego-motion estimation using low-cost mmWave radars via radar velocity factor for pose-graph SLAM,

    Y . S. Park, Y .-S. Shin, J. Kim, and A. Kim, “3D ego-motion estimation using low-cost mmWave radars via radar velocity factor for pose-graph SLAM,”IEEE Robotics and Automation Letters (RA-L), vol. 6, no. 4, pp. 7691–7698, 2021

    2021

  7. [7]

    x-RIO: Radar inertial odometry with multiple radar sensors and yaw aiding,

    C. Doer and G. F. Trommer, “x-RIO: Radar inertial odometry with multiple radar sensors and yaw aiding,”Gyroscopy and Navigation, vol. 12, pp. 329–339, 02 2022

    2022

  8. [8]

    ColoRadar: The direct 3d millimeter wave radar dataset,

    A. Kramer, K. Harlow, C. Williams, and C. Heckman, “ColoRadar: The direct 3d millimeter wave radar dataset,”Intl. J. of Robotics Research (IJRR), vol. 41, no. 4, pp. 351–360, 2022

    2022

  9. [9]

    milliEgo: single- chip mmWave radar aided egomotion estimation via deep sensor fusion,

    C. X. Lu, M. R. U. Saputra, P. Zhao, Y . Almalioglu, P. P. De Gusmao, C. Chen, K. Sun, N. Trigoni, and A. Markham, “milliEgo: single- chip mmWave radar aided egomotion estimation via deep sensor fusion,” inProc. ACM Conf. on Embedded Networked Sensor Systems, Yokohama, JP, Nov. 2020, pp. 109–122. 1/12 2/48 3/108 4/192 #device / #antenna 0.50 0.75 1.00 1.2...

    2020

  10. [10]

    Multi-radar inertial odometry for 3d state estimation using mmWave imaging radar,

    J.-T. Huang, R. Xu, A. Hinduja, and M. Kaess, “Multi-radar inertial odometry for 3d state estimation using mmWave imaging radar,” in Proc. IEEE Intl. Conf. on Robotics and Automation (ICRA), Yoko- hama, Japan, May 2024, pp. 12 006–12 012

    2024

  11. [11]

    Dart: Implicit doppler tomography for radar novel view synthesis,

    T. Huang, J. Miller, A. Prabhakara, T. Jin, T. Laroia, Z. Kolter, and A. Rowe, “Dart: Implicit doppler tomography for radar novel view synthesis,” inProc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 24 118–24 129

    2024

  12. [12]

    Radar fields: Frequency-space neural scene representations for fmcw radar,

    D. Borts, E. Liang, T. Broedermann, A. Ramazzina, S. Walz, E. Pal- ladin, J. Sun, D. Brueggemann, C. Sakaridis, L. Van Gool,et al., “Radar fields: Frequency-space neural scene representations for fmcw radar,” inACM SIGGRAPH 2024 Conference Papers, pp. 1–10

    2024

  13. [13]

    Learned depth estimation of 3D imaging radar for indoor mapping,

    R. Xu, W. Dong, A. Sharma, and M. Kaess, “Learned depth estimation of 3D imaging radar for indoor mapping,” inProc. IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), Kyoto, Japan, Oct. 2022

    2022

  14. [14]

    Towards dense and accurate radar perception via efficient cross-modal diffusion model,

    R. Zhang, D. Xue, Y . Wang, R. Geng, and F. Gao, “Towards dense and accurate radar perception via efficient cross-modal diffusion model,” IEEE Robotics and Automation Letters (RA-L), 2024

    2024

  15. [15]

    Radarocc: Robust 3d occupancy prediction with 4d imaging radar,

    F. Ding, X. Wen, Y . Zhu, Y . Li, and C. X. Lu, “Radarocc: Robust 3d occupancy prediction with 4d imaging radar,”Advances in Neural Information Processing Systems, vol. 37, pp. 101 589–101 617, 2024

    2024

  16. [16]

    Rmap: Millimeter- wave radar mapping through volumetric upsampling,

    A. N. Mopidevi, K. Harlow, and C. Heckman, “Rmap: Millimeter- wave radar mapping through volumetric upsampling,” inProc. IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), abudhabi, UAE, Oct. 2024, pp. 1108–1115

    2024

  17. [17]

    High resolution point clouds from mmwave radar,

    A. Prabhakara, T. Jin, A. Das, G. Bhatt, L. Kumari, E. Soltanaghai, J. Bilmes, S. Kumar, and A. Rowe, “High resolution point clouds from mmwave radar,” inProc. IEEE Intl. Conf. on Robotics and Automation (ICRA), London, UK, June 2023, pp. 4135–4142

    2023

  18. [18]

    Radarsplat: Radar gaussian splatting for high-fidelity data synthesis and 3d reconstruction of autonomous driving scenes,

    P.-C. Kung, S. Harisha, R. Vasudevan, A. Eid, and K. A. Skinner, “Radarsplat: Radar gaussian splatting for high-fidelity data synthesis and 3d reconstruction of autonomous driving scenes,”arXiv preprint arXiv:2506.01379, 2025

    Pith/arXiv arXiv 2025

  19. [19]

    Towards foundational models for single- chip radar,

    T. Huang, A. Prabhakara, C. Chen, J. Karhade, D. Ramanan, M. O’toole, and A. Rowe, “Towards foundational models for single- chip radar,” inProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), October 2025, pp. 24 655–24 665

    2025

  20. [20]

    Development and demonstra- tion of mimo-sar mmwave imaging testbeds,

    M. E. Yanik, D. Wang, and M. Torlak, “Development and demonstra- tion of mimo-sar mmwave imaging testbeds,”IEEE Access, vol. 8, pp. 126 019–126 038, 2020

    2020

  21. [21]

    Probabilistic sar processing for high-resolution mapping us- ing millimeter-wave radar sensors,

    T. Grebner, A. Grathwohl, P. Schoeder, V . Janoudi, and C. Wald- schmidt, “Probabilistic sar processing for high-resolution mapping us- ing millimeter-wave radar sensors,”IEEE Transactions on Aerospace and Electronic Systems, vol. 59, no. 5, pp. 4800–4814, 2023

    2023

  22. [22]

    Radar- based mapping of the environment: Occupancy grid-map versus sar,

    T. Grebner, P. Schoeder, V . Janoudi, and C. Waldschmidt, “Radar- based mapping of the environment: Occupancy grid-map versus sar,” IEEE Microwave and Wireless Components Letters, vol. 32, no. 3, pp. 253–256, 2022

    2022

  23. [23]

    Indoor synthetic aperture radar measurements of point-like targets using a wheeled mo- bile robot,

    Y . E. Ritterbusch, J. Fink, and C. Waldschmidt, “Indoor synthetic aperture radar measurements of point-like targets using a wheeled mo- bile robot,” inEUSAR 2024; 15th European Conference on Synthetic Aperture Radar. VDE, 2024, pp. 595–600

    2024

  24. [24]

    Rio-sar: Synthetic aperture radar imaging of indoor scenes based on radar-inertial odometry using a mobile robot,

    ——, “Rio-sar: Synthetic aperture radar imaging of indoor scenes based on radar-inertial odometry using a mobile robot,”IEEE Trans- actions on Radar Systems, 2024

    2024

  25. [25]

    Simultaneous localization and mapping for indoor mobile robots using synthetic aperture radar images,

    ——, “Simultaneous localization and mapping for indoor mobile robots using synthetic aperture radar images,”IEEE Journal of Indoor and Seamless Positioning and Navigation, 2024

    2024

  26. [26]

    Automotive radar gridmap representa- tions,

    K. Werber, M. Rapp, J. Klappstein, M. Hahn, J. Dickmann, K. Diet- mayer, and C. Waldschmidt, “Automotive radar gridmap representa- tions,” inIEEE MTT-S International Conference on Microwaves for Intelligent Mobility (ICMIM), 2015, pp. 1–4

    2015

  27. [27]

    3d occupancy grid mapping using statistical radar models,

    J. Degerman, T. Pernstål, and K. Alenljung, “3d occupancy grid mapping using statistical radar models,” inIEEE Intelligent V ehicles Symposium (IV), 2016, pp. 902–908

    2016

  28. [28]

    Adap- tions for automotive radar based occupancy gridmaps,

    R. Prophet, H. Stark, M. Hoffmann, C. Sturm, and M. V ossiek, “Adap- tions for automotive radar based occupancy gridmaps,” inIEEE MTT- S International Conference on Microwaves for Intelligent Mobility (ICMIM), 2018, pp. 1–4

    2018

  29. [29]

    Radar-inertial ego-velocity estimation for visually degraded environments,

    A. Kramer, C. Stahoviak, A. Santamaria-Navarro, A.-A. Agha- Mohammadi, and C. Heckman, “Radar-inertial ego-velocity estimation for visually degraded environments,” inProc. IEEE Intl. Conf. on Robotics and Automation (ICRA), Paris, FR, May 2020, pp. 5739– 5746

    2020

  30. [30]

    PreCFAR gridmaps for automotive radar,

    F. Weishaupt, N. Appenrodt, J. F. Tilly, J. Dickmann, and D. Heberling, “PreCFAR gridmaps for automotive radar,” in18th European Radar Conference (EuRAD). IEEE, 2022, pp. 161–164

    2022

  31. [31]

    Realistic sar implementation for automotive applications,

    H. Iqbal, A. Löffler, M. N. Mejdoub, and F. Gruson, “Realistic sar implementation for automotive applications,” inIEEE 17th European Radar Conference (EuRAD), 2021, pp. 306–309

    2021

  32. [32]

    See through smoke: robust indoor map- ping with low-cost mmwave radar,

    C. X. Lu, S. Rosa, P. Zhao, B. Wang, C. Chen, J. A. Stankovic, N. Trigoni, and A. Markham, “See through smoke: robust indoor map- ping with low-cost mmwave radar,” inProc. of the 18th International Conference on Mobile Systems, Applications, and Services, 2020, pp. 14–27

    2020

  33. [33]

    High-resolution image synthesis and semantic manipulation with conditional GANs,

    T.-C. Wang, M.-Y . Liu, J.-Y . Zhu, A. Tao, J. Kautz, and B. Catanzaro, “High-resolution image synthesis and semantic manipulation with conditional GANs,” inProc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2018, pp. 8798–8807

    2018

  34. [34]

    Enabling visual recognition at radio frequency,

    H. Lai, G. Luo, Y . Liu, and M. Zhao, “Enabling visual recognition at radio frequency,” inProceedings of the 30th Annual International Conference on Mobile Computing and Networking (MobiCom), 2024, pp. 388–403

    2024

  35. [35]

    Synthetic-aperture radar processing using fast factorized back-projection,

    L. M. Ulander, H. Hellsten, and G. Stenstrom, “Synthetic-aperture radar processing using fast factorized back-projection,”IEEE Transac- tions on Aerospace and electronic systems, vol. 39, no. 3, pp. 760–776, 2003

    2003

  36. [36]

    J. C. Curlander and R. N. McDonough,Synthetic aperture radar. Wiley, New York, 1991, vol. 11

    1991

  37. [37]

    Focusing of airborne synthetic aperture radar data from highly nonlinear flight tracks,

    O. Frey, C. Magnard, M. Ruegg, and E. Meier, “Focusing of airborne synthetic aperture radar data from highly nonlinear flight tracks,”IEEE transactions on geoscience and remote sensing, vol. 47, no. 6, pp. 1844–1858, 2009

    2009

  38. [38]

    Modeling sar images with a gener- alization of the rayleigh distribution,

    E. E. Kuruoglu and J. Zerubia, “Modeling sar images with a gener- alization of the rayleigh distribution,”IEEE Transactions on Image Processing, vol. 13, no. 4, pp. 527–533, 2004

    2004

  39. [39]

    Fast-lio: A fast, robust lidar-inertial odometry package by tightly-coupled iterated kalman filter,

    W. Xu and F. Zhang, “Fast-lio: A fast, robust lidar-inertial odometry package by tightly-coupled iterated kalman filter,”IEEE Robotics and Automation Letters, vol. 6, no. 2, pp. 3317–3324, 2021

    2021

  40. [40]

    OpenVINS: A research platform for visual-inertial estimation,

    P. Geneva, K. Eckenhoff, W. Lee, Y . Yang, and G. Huang, “OpenVINS: A research platform for visual-inertial estimation,” inProc. IEEE Intl. Conf. on Robotics and Automation (ICRA), Paris, FR, May 2020, pp. 4666–4672

    2020

  41. [41]

    Imaging radar using cascaded mmwave sensor reference design,

    I. Texas Instruments, “Imaging radar using cascaded mmwave sensor reference design,” 2020

    2020