Robot manipulation training data: teleop, sim, or egocentric video? Practitioners weigh in
TendiesTown3 · reddit · 2026-11-22
A developer working on egocentric data collection asks robot-learning practitioners where their manipulation policy / world model training data comes from — teleop, simulation, public video like Ego4D, or paid vendors — and whether human egocentric video actually helps or the embodiment gap kills it. The thread aggregates first-hand training-data practices.
More from Embodied
- Musk: Tesla AI5 chip brings roughly 12X more compute than current in-car hardware — elonmusk · 2026-11-22
- First Drive: Tesla FSD V15 Shines With Reality View Birds-Eye Map and Garage Auto-Parking — Scobleizer · 2026-11-22
- AI drones hunt moths in greenhouses with precision strikes, replacing pesticides — TinfoilTricorn · 2026-11-22
- One photo, 8 VLAs, only 3 correct: ERQA embodied-reasoning benchmark questioned — giffmana · 2026-11-22
- China Installed 59% of New Industrial Robots in 2026 Report, ~354,000 Units — ingliguori · 2026-11-22
- Apple Watch bricks after one shower while a $15 Keep band survives many — oran_ge · 2026-11-22