Glossary
Teleoperation vs egocentric data
Robot-collected demonstrations versus human-worn first-person capture.
Definition
Teleoperation collects demonstrations by a human driving the robot, so data matches the robot’s embodiment exactly but is slow and hardware-bound. Egocentric human capture records people doing tasks directly — faster and more diverse, at the cost of an embodiment gap to close.
01
Why it matters
The two are complementary. Egocentric human data scales coverage cheaply and broadly; teleoperation grounds it in the exact robot. Most programs need a blend rather than one or the other.
02
How Firsthand handles it
Firsthand focuses on high-fitness egocentric human capture — pose, depth and action labels — that pairs with your teleoperation data for embodiment alignment.
Related terms