Dyna-2: Human Data Scaling Law Predictably Improves Zero-Shot Robot Performance
chris_j_paxton · x · 2026-08-11
DynaRobotics shared a key finding from the Dyna-2 research: the human data scaling law implies a scaling law on never-seen robot data.
The research shows that scaling human data for pre-training predictably improves zero-shot offline evaluation on robot data. Their blog also demonstrates that this scaling successfully transfers to on-robot evaluation through post-training on diverse robot embodiments.
More from Embodied
- MemoMind Smart Glasses Exceed Kickstarter Goal 100x, Raise ~$9M — chrisgrayson · 2026-08-11
- Mocking the Embodied AI Boom: Waking Up to 23 Robotics News and 42 Funding Rounds — lukas_m_ziegler · 2026-08-11
- Why Human-Hand Data is the Key to Robot Scaling Laws — JasonMa2020 · 2026-08-11
- Dyna-2: Robot Foundation Model Trained on 1M+ Hours of Human Data Reveals New Scaling Law — JasonMa2020 · 2026-08-11
- University of Tokyo's DRAGON: A Transformable Flying Robot That Acts as a Manipulator — acmoytoy · 2026-08-11
- DynaRobotics Unveils Industry-First Scaling Law Linking Human Video to General Robot Capabilities — JasonMa2020 · 2026-08-11