Intelligent mmWave Sensing for Human-Centric Motor Analysis and Interaction: A Survey

PhD Qualifying Examination


Title: "Intelligent mmWave Sensing for Human-Centric Motor Analysis and
Interaction: A Survey"

by

Mr. Jinjian WANG


Abstract:

Human motion carries rich information about health status, functional ability,
behavioral intention, and interactive commands. With the development of AIoT
systems, motion sensing is moving from controlled laboratory assessment toward
continuous, privacy-sensitive, and daily-life deployment. Existing sensing
modalities, including cameras, wearable sensors, acoustic sensing, and WiFi,
have enabled progress in activity recognition, rehabilitation assessment,
health monitoring, and human-computer interaction, but they still face
limitations in privacy, lighting robustness, wearing burden, long-term
adherence, and deployment flexibility.

Millimeter-wave radar has emerged as a promising contactless modality for
human-centric motor analysis and interaction. Its fine motion sensitivity,
spatial perception capability, lighting robustness, and privacy-preserving
nature make it suitable for capturing both coarse body motion and subtle
physiological or interaction-related movements. Together with advances in
radar signal processing, range-Doppler representation, point-cloud modeling,
phase-based analysis, and deep learning, mmWave sensing is moving beyond
simple motion detection toward fine-grained motor assessment, semantic action
recognition, physiological monitoring, and wearable interaction.

This survey reviews intelligent mmWave sensing from both application and
technical perspectives. We first summarize representative human motion sensing
scenarios and organize sensing problems into coarse motion awareness, motion
parameter and quality analysis, and semantic action recognition with
long-term behavioral understanding. We then discuss sensing modalities,
mmWave advantages, FMCW signal processing, common radar representations,
learning models, robustness, generalization, and technical limitations.
Finally, we present two research attempts, mmBrady for contactless upper-limb
bradykinesia monitoring in Parkinson's disease and RadarGlass for real-time
direction-aware gesture interaction on AR glasses. We conclude by outlining
future directions toward robust, personalized, semantic-aware,
privacy-preserving, and deployable mmWave sensing systems for daily-life
motor analysis and interaction.


Date:                   Monday, 22 June 2026

Time:                   4:00pm - 6:00pm

Venue:                  Room 3494
                        Lifts 25/26

Committee Members:      Prof. Qian Zhang (Supervisor)
                        Prof. Gary Chan (Chairperson)
                        Dr. Xiaomin Ouyang