SCHOLARLY RECORD / ROBOTICS: SCIENCE AND SYSTEMS
Diffusion Policy: Visuomotor Policy Learning via Action Diffusion
Cheng Chi, Zhenjia Xu, Siyuan Feng, et al.
Educational abstract, not a reproduction
This page preserves only an Observatory-written summary and context. The original source remains authoritative for the paper’s language, methods, findings and limitations.
Plain-language summary
The work applies diffusion models to generate robot action sequences from observations.
Why it matters
It introduced a widely studied approach for learning multi-modal manipulation behavior from demonstrations.
Institutions
- Columbia University
- Toyota Research Institute
- MIT