Benjamin Burchfiel

Toyota Research Institute, Brown University, Duke University

Papers

15

Total Citations

642

H-Index

9

About

Benjamin Burchfiel is a leading researcher in robot learning and manipulation, with a focus on developing scalable, data-driven policies that bridge perception and action. His most impactful contribution is **Diffusion Policy**, a groundbreaking framework that reframes visuomotor policy learning as a conditional denoising diffusion process. This work, which has garnered over 338 citations, consistently outperforms existing methods across 15 diverse manipulation benchmarks, establishing a new paradigm for generating robust, high-frequency robot behaviors. Burchfiel also pioneered the **Universal Manipulation Interface**, enabling in-the-wild robot teaching without requiring physical robots at training time (131 citations), and co-developed **OpenVLA**, an open-source vision-language-action model that leverages internet-scale pretraining for fine-tuning robot skills. His research extends to dynamic manipulation of deformable objects through iterative residual policies and to adaptive compliance for force-sensitive tasks. Earlier work on Bayesian Eigenobjects introduced a unified framework for 3D robot perception, combining classification, pose estimation, and geometric completion. With over 600 total citations and a portfolio spanning from foundational perception to cutting-edge diffusion-based control, Burchfiel is shaping the future of generalist robot manipulation.

Research Focus

Key Achievements

9
H-Index
15
Papers
642
Total Citations
43
Avg Citations/Paper
🏆 Most Cited Paper
Diffusion policy: Visuomotor policy learning via action diffusion
338 citations · 2024
📈 Most Prolific Year: 2024 (4 Papers)
🤝 Key Collaborators: 51
🏛 Institutions: Toyota Research Institute, Brown University, Duke University

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 13 days ago