Teaching Foundation Models to Read mmWave: Pose-Guided Kinematic Representation for Human Behavior Understanding

• URL: http://arxiv.org/abs/2608.04127v2
• Date: Sun, 09 Aug 2026 09:08:29 GMT
• Title: Teaching Foundation Models to Read mmWave: Pose-Guided Kinematic Representation for Human Behavior Understanding
• Abstract summary: Millimeter-wave (mmWave) radar provides a privacy-friendly and contactless sensing modality.<n>Existing radar-language methods often rely on synthetic data or lack explicit supervision for human structure and motion dynamics.<n>We present mmMind, a radar-language model that uses a synchronized 3 pose training model.
• Score: 23.73222950886392
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language model agents need to perceive human behavior in physical environments. Millimeter-wave (mmWave) radar provides a privacy-friendly and contactless sensing modality, but radar observations are difficult to align with language. Existing radar-language methods often rely on synthetic data or lack explicit supervision for human body structure and motion. We present mmMind, a radar-language model that uses synchronized 3D pose as training-only supervision. A spatio-temporal radar encoder is pretrained to capture body configuration and motion dynamics, after which the pose head is removed so that inference requires radar alone. The learned radar representations are then aligned with an LLM for behavior captioning and spatio-temporal question answering. We also introduce mmMind-Bench, a real-world mmWave-language benchmark containing 17.9 hours of recordings from 23 participants across seven indoor environments. Experiments on captioning, question answering, and unseen-action generalization show that mmMind consistently outperforms existing radar-language baselines, while ablations confirm the importance of pose-guided pretraining.

Related papers

• Talk2Sensors: 3D Visual Grounding in Autonomous Driving via Sensor-Adaptive Physical Cue Matching [66.91228384649921]
  We introduce Talk2Sensors, the first multi-sensor 3D visual grounding dataset built upon camera, LiDAR, and 4D radar.<n>We propose TSFormer, a unified Transformer-based framework for language-guided 3D visual grounding in autonomous driving.
  arXiv  Detail & Related papers  (2026-08-05T07:58:35Z)
• Wave2Body: Rethinking mmWave Human Pose Estimation as Radar-to-Body Token Translation [17.79621154844076]
  Millimeter-wave (mmWave) radar enables privacy-friendly human sensing, but its sparse point clouds are physical measurements of view-dependent electromagnetic reflections and only indirectly characterize body articulation.<n>Existing methods directly regress joint coordinates from paired radar-pose data, relying on the same limited paired supervision to learn radar perception, human-body structure, and their alignment.<n>We propose Wave2Body, a radar-to-body token translation framework that decouples these learning targets using a self-supervised mmWave tokenizer, a pretrained compositional body tokenizer that defines the output space, and a lightweight translator between them.
  arXiv  Detail & Related papers  (2026-07-21T09:06:31Z)
• RadarTwin: Scene-Specific mmWave Radar Simulation and Learning for Mobile Indoor Perception [2.0255467447384015]
  RadarTwin is a framework for generating deployment-specific radar training data before real data collection.<n>We show that simulated and real radar share the same object-discriminative shape and material features.<n>A representation trained on simulation alone recognizes real objects at 2.5 times chance with no real radar labels, and a few labeled examples raise this to 95.3% on a 12-way recognition task.
  arXiv  Detail & Related papers  (2026-06-24T05:50:26Z)
• RadarGen: Automotive Radar Point Cloud Generation from Cameras [64.69976771710057]
  We present RadarGen, a diffusion model for synthesizing realistic automotive radar point clouds from multi-view camera imagery.<n>RadarGen adapts efficient image-latent diffusion to the radar domain by representing radar measurements in bird's-eye-view form.<n>We show that RadarGen captures characteristic radar measurement distributions and reduces the gap to perception models trained on real data.
  arXiv  Detail & Related papers  (2025-12-19T18:57:33Z)
• RadarLLM: Empowering Large Language Models to Understand Human Motion from Millimeter-wave Point Cloud Sequence [10.115852646162843]
  We present Radar-LLM, the first framework that leverages large language models (LLMs) for human understanding using millimeter-wave radar as the sensing modality.<n>To address data scarcity, we introduce a physics-aware pipeline synthesis that generates realistic radar-text pairs from motion-text datasets.<n>Radar-LLM achieves state-of-the-art performance across both synthetic and real-world benchmarks, enabling accurate translation of millimeter-wave signals to natural language descriptions.
  arXiv  Detail & Related papers  (2025-04-14T04:18:25Z)
• RobuRCDet: Enhancing Robustness of Radar-Camera Fusion in Bird's Eye View for 3D Object Detection [68.99784784185019]
  Poor lighting or adverse weather conditions degrade camera performance.<n>Radar suffers from noise and positional ambiguity.<n>We propose RobuRCDet, a robust object detection model in BEV.
  arXiv  Detail & Related papers  (2025-02-18T17:17:38Z)
• Talk2Radar: Bridging Natural Language with 4D mmWave Radar for 3D Referring Expression Comprehension [21.598751853520834]
  4D millimeter-wave radars provide denser point clouds than conventional radars and perceive both semantic and physical characteristics of objects.<n>To foster the development of natural language-driven context understanding in radar scenes for 3D visual grounding, we construct the first dataset, Talk2Radar.<n>We propose a novel model, T-RadarNet, for 3D Referring Expression on point clouds, achieving State-Of-The-Art (SOTA) performance on the Talk2Radar dataset.
  arXiv  Detail & Related papers  (2024-05-21T14:26:36Z)
• Radar Fields: Frequency-Space Neural Scene Representations for FMCW Radar [62.51065633674272]
  We introduce Radar Fields - a neural scene reconstruction method designed for active radar imagers. Our approach unites an explicit, physics-informed sensor model with an implicit neural geometry and reflectance model to directly synthesize raw radar measurements. We validate the effectiveness of the method across diverse outdoor scenarios, including urban scenes with dense vehicles and infrastructure.
  arXiv  Detail & Related papers  (2024-05-07T20:44:48Z)
• Language-Guided 3D Object Detection in Point Cloud for Autonomous Driving [91.91552963872596]
  We propose a new multi-modal visual grounding task, termed LiDAR Grounding. It jointly learns the LiDAR-based object detector with the language features and predicts the targeted region directly from the detector. Our work offers a deeper insight into the LiDAR-based grounding task and we expect it presents a promising direction for the autonomous driving community.
  arXiv  Detail & Related papers  (2023-05-25T06:22:10Z)
• RadarLoc: Learning to Relocalize in FMCW Radar [36.68888832365474]
  We propose a novel end-to-end neural network with self-attention, termed RadarLoc, which is able to estimate 6-DoF global poses directly. We validate our approach on the recently released challenging outdoor dataset Oxford Radar RobotCar.
  arXiv  Detail & Related papers  (2021-03-22T03:22:37Z)
• RadarNet: Exploiting Radar for Robust Perception of Dynamic Objects [73.80316195652493]
  We tackle the problem of exploiting Radar for perception in the context of self-driving cars. We propose a new solution that exploits both LiDAR and Radar sensors for perception. Our approach, dubbed RadarNet, features a voxel-based early fusion and an attention-based late fusion.
  arXiv  Detail & Related papers  (2020-07-28T17:15:02Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.