Physical AI

Sense the physical world. Build machine understanding.

Spectrum · Polarization · 3D

Integrated computation and perception Understand real environments

Front-end computation and multimodal perception work together
to support robotic recognition and manipulation.

Transparent / Translucent

Inform boundary and surface estimation

Reflective / Mirror-like

Polarization adds surface cues

Same color, different materials

Spectra help distinguish materials

Low-texture surfaces

Depth adds geometric cues

VISUAL FOUNDATION MODEL

VFM Visual Foundation Model

Shared visual representations for diverse tasks

01 Image input

A skier in an orange jacket on a snowy mountain slope

Visual task input

02 Visual representation

VFM

Shared visual understanding

03 Task adaptation and outputs

A skier in an orange jacket on a snowy mountain slopeSkiing
Image classificationClassification
A skier in an orange jacket on a snowy mountain slopeSkier
Object detectionDetection
Illustrative segmentation of the same ski scene, separating the skier from the background
Image segmentationSegmentation
Illustrative depth estimation of the same ski scene, showing distances between the skier, slope, and mountains
Depth estimationDepth

Image and task outputs are conceptual illustrations

CONTACT

Together
make innovation happen

Discuss your application, technical needs, and opportunities for joint validation.

Discuss a partnership

By submitting, you agree that we may use these details to contact you.