Sensor-fusion work shared with Munich collaborators.

A first internal note on vision–force–inertial fusion has been circulated to our research collaborators in Munich. The aim is simple to state and hard to do: one coherent world model the controller can trust, rather than three streams argued over at runtime.

The note covers the early work — time alignment across the three streams, a small study on what the controller actually needs from each, and where the model breaks first when the bench stops looking like the bench. It is work in progress, shared early on purpose.

Sharing early is the point of the platform. The faster a research idea meets the hardware, the faster it either becomes a deployment or gets retired — and the next experiment gets the budget.