Research

Dexterous skill emerges from the precise coordination of movement, force, timing, and feedback learned through years of practice.

Surgery is a powerful example: subtle hand movements and finely controlled forces are used to manipulate tools in ways that are difficult to measure, model, or fully understand. By capturing these hidden dynamics, we can translate human expertise into new ways to teach complex skills and train autonomous systems, with applications ranging from surgical robotics to advanced laboratory automation.

Measuring hand movement and force continuously makes it possible to drive machines directly.

Robotic teleoperation is a compelling example: movements and forces decoded at the wrist can drive a robotic arm in real time, with the full expressiveness of the human operator and without bulky gloves or cameras. The same principle extends to any system that benefits from precise, continuous human input.

As digital systems become easier to automate and impersonate, knowing when a real person is present and acting becomes increasingly important. Each person's neuromotor activity- and the movements it produces— carries a unique signature, shaped by intrinsic biophysics and a lifetime of motor learning.

These approaches can continuously verify identity and passively watermark human actions as they occur. Working with ARPA-H, we are applying this approach to healthcare settings where rapid, continuous authentication can improve security and accountability without slowing down urgent clinical work.