RAGent: Physics-Aware Agentic Reasoning for Training-Free mmWave Human Activity Recognition

TL;DR

RAGent is a training-free mmWave radar-based human activity recognition framework that uses evidence-grounded inference and a reusable radar knowledge base, enabling scalable deployment without retraining.

Contribution

It introduces a novel deployment-time inference approach that eliminates the need for retraining or adaptation in mmWave HAR systems.

Findings

Achieves 93.39% accuracy without domain-specific retraining.

Generalizes robustly across different deployment domains.

Constructs a reusable radar knowledge base via cross-modal supervision.

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-time training-free framework for mmWave HAR that reformulates recognition as evidence-grounded inference over reusable radar knowledge rather than deployment-specific model optimization. Offline, RAGent constructs a reusable radar knowledge base through constrained cross-modal supervision, where a Vision-Language Model (VLM) transfers activity semantics from synchronized videos to paired radar segments without manual radar annotation. At deployment time, RAGent recognizes activities from radar alone by retrieving physically comparable precedents in an explicit kinematic space and resolving the final label through structured multi-role reasoning. The reasoning protocol is further refined offline through zero-gradient self-evolution. Extensive experiments on a self-collected dataset show that RAGent achieves 93.39% accuracy without per-domain retraining or target-domain adaptation, while generalizing robustly across domains.