OVISION
Purpose-built hardware, capturing everything vision can give.
O-VISION: An egocentric camera system that captures what the demonstrator sees: object perception, localization, 3D coordinates, and hand pose, all on a single calibrated timeline.
Cameras
3 · 1 perception + 2 CV
Frame rate
60 fps
Time sync
Sub-ms · single calibrated timeline
Calibration
Factory · intrinsics, extrinsics, distortion
Software stack for OVISION data.
Every OVISION recording is refined into training-ready robot data — segmented, reconstructed, and retargeted into action trajectories that transfer directly to real and simulated policies.
Objects and hands interactions
High-fidelity 3D depth
One system, any form factor.
Move the same sensor stack and algorithms into whatever the job needs — headset, safety glasses, or a custom rig. The capture pipeline stays the same.
Headset. Full sensor stack in a lab-grade rig.
Safety glasses. Job-site shell, same capture pipeline.
Custom rig. Fit the machine — the algorithms stay the same.