Purpose-built, multi-sensor capture.
We produce synchronized, multi-sensor data to translate the real world into useful training signals. We capture with human-worn sensors, UMI-style setups, and tele-operated devices, with sensors configured from simple RGB + IMU to depth, tactile, and more. The physical world doesn't fit one rig: we work with you to tailor capture modalities where the task demands it.
Egocentric human capture, at scale
Lightweight, head-mounted capture that runs in real homes and workplaces – synchronized video, motion, and depth from the human point of view.
Multi-modal and richly annotated
We work with you to ensure sensors and post-processing will drive downstream performance in models – not data for data’s sake. We support many modalities of capture and annotation approaches, tailored to your goals.
