Research

We develop sensing systems that reveal environmental phenomena that are difficult for people and sensors to directly perceive in everyday settings.

• Repurposing commodity sensors to capture subtle airflow, temperature-field, and soil-state changes.
• Turning low-cost platforms into deployable tools for environmental understanding in real-world scenarios.

We build practical health sensing systems on consumer wearables and mobile devices for robust monitoring in everyday environments.

• Repurposing earbuds, phones, and lightweight wearable platforms for continuous physiological and behavioral sensing.
• Designing sensing pipelines that remain robust under motion, sparse sensing, and real-world deployment constraints.

We study how ambient and wearable sensing systems can capture fine-grained human activities and support higher-level reasoning about daily behaviors in real-world environments.

• Combining multi-modal sensing for detailed activity understanding across body movement, context, and daily-life interactions.
• Building sensing pipelines that support both precise recognition and richer behavioral reasoning.

This direction explores how low-cost sensing and actuation pipelines can give robots richer tactile perception and more practical interaction capabilities.

We study how AI can help mobile and sensing systems make better decisions about communication, adaptation, and resource management under real-world constraints.

• Using foundation and generative models to improve system efficiency, communication, and deployment practicality.
• Designing adaptive mobile systems under limited bandwidth, compute, and storage budgets.

We develop both attacks and defenses to better understand the trust, privacy, and security boundaries of sensing systems.

• Studying how sensing modalities can leak private information or be manipulated by adversaries.
• Designing practical defenses for mobile, acoustic, RF, and AI-driven sensing platforms.

This direction explores mobile systems that use sensing to support more adaptive and context-aware native AI capabilities on personal devices.

This area studies how AI models and sensing pipelines can co-adapt to make mobile sensing systems more robust, data-efficient, and deployable.

This project studies wearable AI systems that can capture, organize, and retrieve important personal experiences to support human memory in everyday life.