t("re.hero.eyebrow")
t("re.hero.lede")
What the institute studies
A model that writes text can be wrong and waste a moment. A machine holding a tool cannot.
Physical AI works under gravity, contact, noise and a clock that does not stop. It is a different problem from the AI most people have encountered, and it fails differently.
Everything the institute builds is published in the open and free to run. Simulations run in an ordinary browser on a student's own device: no licence, no install and no data centre, which matters wherever equipment budgets are thin.
Every figure above is checkable at physicalai-bmi.org. Publications carry a verification grade, and results the institute has not independently confirmed are marked as such.
Three laboratories
Airspace integration, aviation training, and the STEM workforce.
The Glass Lab's subject is the same subject as the institute's aviation programmes, which is why the research and the teaching are held in one institution.
Uncrewed aircraft in the field.
Logistics, search and rescue, last-mile delivery and coastal surveillance. Aircraft doing work, in conditions that do not cooperate.
Provable, physically grounded AI.
Intelligence that runs on the device rather than in a data centre, and can show why its next action is safe before it takes it.
Flagship work
Optical rays carve empty space and stop at surfaces; spatial RF penetrates the barrier to resolve the shadow.
Where a system is, and what is around it, treated not as an internal picture it carries but as something a mesh of optical and radio-frequency nodes resolves out of a volume, into occupied, empty and unknown.
Read the paper
The digital twin used in coursework and live experiments.
US agriculture is counted well at the top and thinly where a machine has to work. This review sizes what physical AI can take on across the estate, the labour market and the logistics layer. Turning over the US machine fleet through new sales alone would take about forty-six years.
Read the paperPerception and world models
Depth, detection, multimodal fusion.
Manipulation and dexterity
Contact-rich control and tactile sensing.
Learning and sim-to-real
Train in simulation. Survive the real world.
Why a teaching institute runs a research institute
Course material in a fast-moving field goes stale quickly, and buying it in means teaching last year's practice.
The institute writes its own. Research done in the labs becomes the curriculum we teach, and the teaching sends problems back the other way.
Papers, courses, instruments and the live labs are all at the institute itself.