Charles Schaff
Papers
11
Total Citations
81
H-Index
5
About
Charles Schaff’s research lies at the intersection of dexterous manipulation, soft robotics, and the co-optimization of robot design and control. His major contributions include pioneering frameworks that jointly optimize a robot’s physical structure and its learned policy using deep reinforcement learning, demonstrating that design and control are inherently coupled. Notably, his work on soft legged robots shows how compliance provides a form of “mechanical intelligence,” enabling passive behaviors that simplify control—a concept he validated through sim-to-real transfer. In dexterous manipulation, Schaff led the winning submission to the Real Robot Challenge, combining grasp and motion planning on the TriFinger platform. He also co-developed a remotely accessible robot cluster to enable reproducible dexterous manipulation research. His most cited papers (24 and 21 citations) benchmark structured policies for real-world dexterous tasks and co-optimize soft robot crawlers, while his earlier work on beacon-based localization for GPS-denied environments laid groundwork for autonomous navigation. Schaff’s research consistently bridges simulation and reality, advancing data-driven methods that make complex robotic behaviors practical and reproducible.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Jointly optimizing placement and inference for beacon-based localization8 citations · 2017
- 4
- 5
- 6Residual Policy Learning for Shared Autonomy3 citations · 2020
- 7A Robot Cluster for Reproducible Research in Dexterous Manipulation3 citations · 2021
- 8Sim-to-real transfer of co-optimized soft robot crawlers2 citations · 2023
- 9Jointly Optimizing Placement and Inference for Beacon-based Localization2 citations · 2017
- 10