Physical AI · Data infrastructure

Human skill, machine readable.

Vision, motion, depth and language from everyday human activity. Structured for embodied systems.

RGB · Depth · SLAM · Hand pose · Expert labels

Talk to usApply as an expert
Pipeline output

Deployment-grade multi-modal data

Aligned at the hardware level. Verified by expert reviewers before delivery.

Capabilities

Eight layers from one capture

Six layers come off the sensor rig. Two are written and checked by people.

RGB Video

RGB

Egocentric · multi-view

Hand Pose

HND

2D + 3D · MANO

Head Pose

HED

6-DoF · world-aligned

SLAM

MAP

Trajectory + map

Depth

DPH

Metric · RGB-registered

Object Tracking

TRK

Persistent IDs

Task Labels

TSK

Segment-level

Action Labels

ACT

Frame-level

Get in touch

Start with your task

Take the existing dataset, or commission new capture.

Talk to us