Dex-One2Many: One Human Video Trains Multi-Morphology Robot Policies

Dex-One2Many uses a Real2Sim2Real pipeline with neural-symbolic abstraction to train dexterous manipulation policies from a single human video, enabling zero-shot transfer to real robots across morphologies and object configurations.

2026-10-09 ~ 2026-10-10 · 2 related posts

• Dex-One2Many: one human video trains a dexterous hand that transfers zero-shot to real robots — furongh · 2026-10-09
• Dex-One2Many: Real2Sim2Real Turns One Human Video Into Robots Generalizing Across Configurations — furongh · 2026-10-10