Siyuan Feng

About

Siyuan Feng is a robotics researcher whose work spans humanoid robot control, optimization-based motion planning, and robot learning — fields where his contributions have garnered over 1,300 citations across a decade of influential publications. He rose to early prominence through his foundational work on full-body control of the Boston Dynamics Atlas robot, developing inverse dynamics and optimization-based frameworks that enabled robust, compliant humanoid locomotion and manipulation. This work played a central role in multiple teams' approaches to the landmark DARPA Robotics Challenge (2013–2015), where his algorithms helped achieve reliable, fall-free performance under real-world disaster-response conditions — an achievement documented across several highly cited papers. His receding-horizon footstep optimization work further advanced dynamic walking robustness in complex environments. More recently, Feng has made a significant pivot into robot learning, co-authoring the widely adopted Diffusion Policy framework (338 citations), which reformulates visuomotor robot control as a conditional diffusion process and has rapidly become a cornerstone reference in imitation learning. His Universal Manipulation Interface work extends this vision by enabling flexible, in-the-wild robot teaching. Together, these contributions mark Feng as a researcher who has shaped both classical and modern paradigms in robot autonomy.

Research Focus

Key Achievements

Top Papers

1. 1
   Diffusion policy: Visuomotor policy learning via action diffusion

   338 citations · 2024
2. 2
   Optimization‐based Full Body Control for the DARPA Robotics Challenge

   229 citations · 2015
3. 3
   Optimization based full body control for the atlas robot

   163 citations · 2014
4. 4
   Universal Manipulation Interface: In-The-Wild Robot Teaching Without In-The-Wild Robots

   131 citations · 2024
5. 5
   Optimization based controller design and implementation for the Atlas robot in the DARPA Robotics Challenge Finals

   107 citations · 2015
6. 6
   No falls, no resets: Reliable humanoid behavior in the DARPA robotics challenge

   105 citations · 2015
7. 7
   Robust dynamic walking using online foot step optimization

   91 citations · 2016
8. 8
   3D walking based on online optimization

   83 citations · 2013
9. 9
   Human‐in‐the‐loop Control of a Humanoid Robot for Disaster Response: A Report from the DARPA Robotics Challenge Trials

   78 citations · 2015
10. 10
    What Happened at the DARPA Robotics Challenge Finals

    57 citations · 2018

Key Collaborators

Contact & Links