SCORE: Support-Constrained RL Enables Real-World Policy Improvement without Real-World Experience
Preprint, 2026
Website arXiv soon
Raymond Yu
Robot learning researcher
Hey there! I'm an undergraduate studying computer science at the University of Washington, advised by Abhishek Gupta in the WEIRD Lab.
My research focuses on teaching robots to learn from all kinds of data, including suboptimal demonstrations, vision data from the internet, and raw video, rather than relying on large amounts of clean, hand-collected robot data. I'm interested in systems that can squeeze useful behavior out of cheap, abundant, and imperfect data sources.
I'm applying for PhD programs beginning this fall, and would love to chat about research fit or potential collaborations.
News
- Aug 2025Honored to receive the 2025 Microsoft Endowed Scholarship.
- Jun 2025Unified World Models won the Best Paper Award at the ICML 2025 World Modeling Workshop.
- Apr 2025Unified World Models accepted to RSS 2025.
- Jan 2025DRAWER accepted to CVPR 2025; HAMSTER accepted to ICLR 2025.
Publications
* denotes equal contribution
