The opportunity

Most language-steered motion systems describe where a limb should go. Fewer describe what happens when it meets something. Atomic Motion offers a compact control vocabulary—signed translation, rotation, and hold atoms—while Beyond Gestures shows that a wrist pressure array can recover hand pose and distributed force. Together they suggest a tractable next step: contact-conditioned motion language.

A small experiment

Build a three-layer prototype. First, encode a short interaction as atomic commands plus a contact state: approach, touch, load, slide, release. Second, feed the command stream and wrist pressure features to a masked motion generator. Third, decode both motion and a plain-language explanation of the contact event. Keep the first benchmark narrow: three objects, three force levels, and two users.

The evaluation should separate kinematics from interaction. Measure joint error and timing, but also contact onset error, peak-force error, release timing, and whether the generated hand remains physically plausible. Ask a human rater a separate question: does the movement communicate the intended action without overclaiming force it did not sense?

Why this is timely

The design joins two kinds of supervision that are usually separated. Atomic commands are legible and easy to edit; pressure is messy but grounded in what the body actually felt. A model that predicts both can expose disagreement instead of smoothing it away. That is useful for assistive interfaces, teleoperation, and creative tools where “almost the same gesture” can mean a different interaction.

The project should remain deliberately modest. The cited pressure results are single-user and the manipulation study reports a limited number of real trials. The first success criterion is not a new benchmark record; it is a reproducible failure report showing when an atomic command and a sensed contact state diverge.

This proposal extends our community report on contact and shape and the research digest on structured motion. It also offers a concrete bridge to the month’s synthesis: representation is not only a storage choice, but a theory of what counts as an event.

References