RAGent enables training-free, deployment-time human activity recognition for mmWave radar using agentic reasoning.
March 31, 2026
Original Paper
RAGent: Physics-Aware Agentic Reasoning for Training-Free mmWave Human Activity Recognition
arXiv · 2603.27571
The Takeaway
It eliminates the 'collect-tune-redeploy' cycle for radar sensors by using VLMs to transfer knowledge to radar segments offline, then performing evidence-grounded inference at runtime. It achieves 93% accuracy in new domains without any target-domain training data.
From the abstract
Millimeter-wave (mmWave) radar enables privacy-preserving human activity recognition (HAR), yet real-world deployment remains hindered by costly annotation and poor transferability under domain shift. Although prior efforts partially alleviate these challenges, most still require retraining or adaptation for each new deployment setting. This keeps mmWave HAR in a repeated collect-tune-redeploy cycle, making scalable real-world deployment difficult. In this paper, we present RAGent, a deployment-