RESEARCH OVERVIEW
Sensing through
interaction traces.
Every interaction between a subject and its environment leaves a trace. A footstep vibrates a floor. A resting person changes a bed’s response to music. A swimming animal leaves a wake.
My work reveals hidden traces through active probing, enhances weak traces through sensor design, and interprets existing traces through physics-informed learning. These methods connect physical interactions to information about people and animals.
Reveal hidden traces
Enhance weak traces
Interpret existing traces
Research projects
01 Reveal hidden traces
Body weight estimation
Music-induced bed vibrations reveal body weight.
02 Enhance weak traces
Fish monitoring
Whisker design makes fish wakes stand out from background flow.
03 Interpret existing traces
Emotion recognition
Footstep-induced floor vibrations reveal emotion, with learning personalized by gait similarity.
Try it here.
Collapse −Expand +
Research approaches
A person rests on a bed.
Body weight changes the mechanics of the coupled body–bed system.
Music excites the bed.
Music-induced vibrations carry traces of how the person’s body changes the bed’s response.
Vibration sensors record the response.
Vibration sensors on the bed frame measure how the vibrations propagate through the bed–body system.
The model estimates body weight.
The model links these interaction traces to body weight using structural dynamics and physics-informed learning.
REVEAL HIDDEN TRACES
Music excites the bed, making the body’s effect on vibration transmission measurable. Our system interprets this interaction trace through a transfer function and a neural network guided by bed–body structural dynamics.