mlsanityData to 14 Oct 2026

DiFF: Doppler-informed Flow Matching for Human Motion Flow

arXiv 2609.39098 ↗First posted 30 Sep 2026cs.RO cs.CV cs.HCcited by 0

On the map: Diffusion models → Diffusion and flow policies for robotics also fits: 3D, point cloud and scene generation

Abstract (arXiv)

Perceiving human motion via privacy-preserving 4D millimeter-wave (mmWave) radar is critical for next-generation human-robot interaction (HRI), where point cloud scene flow serves as a foundational motion representation. Yet the extreme sparsity and noise of 4D radar point clouds make non-rigid motion flow estimation severely ill-posed--a challenge that existing rigid-centric methods and prior works fail to adequately address, largely because they neglect the rich Doppler velocity cues inherent in 4D radar. We propose DiFF, a generative framework that marries Doppler-informed motion priors with a Kolmogorov-Arnold Network (KAN)-based conditional flow matching model. At its core, a KAN-attention mechanism enables expressive feature extraction, while a prior-guided generative process harnesses Doppler cues to regularize the ill-posed solution space. Extensive experiments show that DiFF achieves state-of-the-art (SOTA) performance across diverse real-world datasets, reducing 3D endpoint error to the millimeter scale on the mmBody benchmark.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models. Builds on 33 of its 47 references that are in the map. First posted on arXiv 30 Sep 2026.

Builds on

33 of 47 references.

References · 33

  1. 2022Flow Matching for Generative Modelingcited by 5,702
  2. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
  3. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  4. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,700
  5. 2024KAN: Kolmogorov-Arnold Networkscited by 649
  6. 2018FlowNet3D: Learning Scene Flow in 3D Point Cloudscited by 242
  7. 2018CodeSLAM - Learning a Compact, Optimisable Representation for Dense Visual SLAMcited by 152
  8. 2020FLOT: Scene Flow on Point Clouds Guided by Optimal Transportcited by 129
  9. 2019Just Go with the Flow: Self-Supervised Scene Flow Estimationcited by 98
  10. 2020PV-RAFT: Point-Voxel Correlation Fields for Scene Flow Estimation of Point Cloudscited by 74
  11. 2020FlowStep3D: Model Unrolling for Self-Supervised Scene Flow Estimationcited by 71
  12. 2020Self-Supervised Monocular Scene Flow Estimationcited by 64
21 more
  1. 2021CamLiFlow: Bidirectional Camera-LiDAR Fusion for Joint Optical Flow and Scene Flow Estimationcited by 56
  2. 2023Hidden Gems: 4D Radar Scene Flow Learning Using Cross-Modal Supervisioncited by 36
  3. 2022mmBody Benchmark: 3D Body Reconstruction Dataset and Analysis for Millimeter Wave Radarcited by 32
  4. 2022Self-Supervised Scene Flow Estimation with 4-D Automotive Radarcited by 29
  5. 2023RPEFlow: Multimodal Fusion of RGB-PointCloud-Event for Joint Optical Flow and Scene Flow Estimationcited by 26
  6. 2023milliFlow: Scene Flow Estimation on mmWave Radar Point Cloud for Human Motion Sensingcited by 20
  7. 2023A New Wave in Robotics: Survey on Recent mmWave Radar Applications in Roboticscited by 18
  8. 2024BlinkVision: A Benchmark for Optical Flow, Scene Flow and Point Tracking Estimation using RGB Frames and Eventscited by 13
  9. 2022DICP: Doppler Iterative Closest Point Algorithmcited by 8
  10. 2022Picking Up Speed: Continuous-Time Lidar-Only Odometry using Doppler Velocity Measurementscited by 6
  11. 2024DiffSF: Diffusion Models for Scene Flow Estimationcited by 6
  12. 2022Learning Moving-Object Tracking with FMCW LiDARcited by 5
  13. 2024Millimeter Wave Radar-based Human Activity Recognition for Healthcare Monitoring Robotcited by 5
  14. 2023Need for Speed: Fast Correspondence-Free Lidar-Inertial Odometry Using Doppler Velocitycited by 4
  15. 2024Generalizable Indoor Human Activity Recognition Method Based on Micro-Doppler Corner Point Cloud and Dynamic Graph Learningcited by 4
  16. 2025DoGFlow: Self-Supervised LiDAR Scene Flow via Cross-Modal Doppler Guidancecited by 4
  17. 2019Automatic Arm Motion Recognition Based on Radar Micro-Doppler Signature Envelopescited by 3
  18. 2025TARS: Traffic-Aware Radar Scene Flow Estimationcited by 3
  19. 2025KAN or MLP? Point Cloud Shows the Way Forwardcited by 3
  20. 2021MonoPLFlowNet: Permutohedral Lattice FlowNet for Real-Scale 3D Scene FlowEstimation with Monocular Imagescited by 2
  21. 2026FMCW-LIO: A Doppler LiDAR-Inertial Odometrycited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025SDDiff: Boost Radar Perception via Spatial-Doppler Diffusion
  2. 2025FlowMotion: Target-Predictive Conditional Flow Matching for Jitter-Reduced Text-Driven Human Motion Generation
  3. 2026You Only Flow Once: Calibrated and Real-Time Radar Pose Estimation with Multi-Hypothesis Normalizing Flows
  4. 2026MotionSpaceFlow: Representation-Aware Flow Matching in Direct Motion Space
  5. 2026mmHRI: Towards Privacy-Preserving Human-Robot Interaction with Millimeter-Wave Radar
  6. 2025Autoregressive Flow Matching for Motion Prediction
  7. 2026Frequency-Conditioned Flow Matching for Vision-Language-Action Models
  8. 2026Privacy-Preserving Full-Body Meshing from mmWave Radar via Mesh Foundation Model Supervision