Huanqi Yang

Huanqi Yang

Ph.D., Computer Science, City University of Hong Kong
huanqi.yang@my.cityu.edu.hk

I work on embodied AI and physical agents: AI that leaves the screen, perceives the physical world, and acts on it through real devices. I am based in Hong Kong. My goal is a physical agent that can sense its surroundings, protect what it senses, and reprogram the hardware it runs on, closing the loop between perception and action without a human engineer in the middle.

I have built each part of that loop. For sensing, I use radio and wearable signals to recognize gait, gestures, handwriting and health, without cameras. For security, I generate encryption keys directly from the wireless channel for LoRa and vehicle networks. For action, AutoEmbed (SenSys 2026) is an LLM agent that writes, uploads and debugs firmware on real embedded boards, so the agent can deploy new sensing and security functions by itself.

This work has received the Mark Weiser Best Paper Award at IEEE PerCom 2023, the Best Paper Award at ACM EIFCom 2024, and a cover article in IEEE JBHI. I received my Ph.D. from City University of Hong Kong, advised by Prof. Weitao Xu.

Research

Wireless & wearable sensing

Generative and cross-modal models that turn radio and IMU signals into gait, gesture and health information. XGait, iRadar, EMGSense.

Physical-layer security

Key generation and authentication from the wireless channel for IoT and vehicles. ChirpKey, Vehicle-Key, Wave-for-Safe.

Agents for embedded systems

LLM agents that program IoT hardware, and efficient LLM inference on edge devices. AutoEmbed, E²LLM.

Selected projects

Full publication list