We investigate how intelligence moves across bodies, from human sensorimotor experience to robots and back through multimodal feedback.
Different bodies may share common physical intelligence. We seek embodiment-invariant representations of motion, contact, force, object interaction, and intent that enable sensorimotor knowledge to transfer across humans and robots. The loop is closed through tactile, haptic, visual, and physical feedback, allowing humans and robots to co-adapt rather than rely on one-way teleoperation.
We develop soft, wearable, and multimodal interfaces that capture whole-body motion, hand motion, contact, force, and physical state without constraining natural behavior.
We study cross-embodiment representations that preserve task-relevant motion and contact information even when human and robot morphologies, sensing spaces, and actuation spaces differ.
We integrate tactile perception, compliant mechanisms, dexterous manipulation, whole-body control, and humanoid platforms to realize intelligence through physical interaction.
We move beyond one-way demonstration toward bidirectional co-adaptation: robots learn from human intervention and sensorimotor experience, while humans receive meaningful feedback from robotic interaction.
Physical avatar systems, dexterous interfaces, artificial skin, whole-body control, and humanoid platforms for transferring human action into robot bodies.
Soft sensors, wearable systems, variable-impedance components, and haptic interfaces for capturing and returning human physical experience.
Multimodal representation learning, imitation learning, visuo-tactile policies, and shared autonomy for transferring and refining sensorimotor skills across human and robot bodies.
BiRC Lab