← Use cases
Use case · Embodied AI
Egocentric data for embodied AI.
The grounded, first-person perception stream agents need to learn to see, move, and act in the physical world.
The need
Embodied agents perceive and act from a first-person viewpoint. Training their perception and planning requires long-horizon, multimodal egocentric experience — video, ego-motion, and gaze — captured as a person actually moves through and interacts with real environments.
What we provide
- Long-horizon, everyday-activity footage
- Navigation and object-interaction sequences
- Synced gaze, IMU, depth, and ego-motion
- Wide environment and scenario diversity
How it's collected
- Daily-living and goal-directed task scenarios
- Consented contributors across varied settings
- Multimodal head-mounted capture
How it's delivered
- Temporally segmented and annotated sequences
- Model-ready formats
- Auditable consent and licensing
More use cases
Need Embodied AI data?
Tell us the tasks and environments — we'll scope the capture.