Zichen Jeff Cui

Papers

Total Citations

H-Index

About

Zichen Jeff Cui is a robotics and machine learning researcher whose work sits at the intersection of imitation learning, sequential decision-making, and large-scale behavior modeling. His research addresses one of the most pressing challenges in robot learning: how to effectively leverage diverse, uncurated human demonstration data to train generalizable robot policies. Cui's most recognized contribution, "Behavior Transformers: Cloning k modes with one stone" (2022, 31 citations), tackles the fundamental problem of multi-modal behavior in human demonstrations. By adapting transformer-based architectures — proven so powerful in language and vision — to capture the full distribution of human behaviors rather than collapsing to a single mode, his work represents a meaningful step toward closing the gap between robot learning and its more data-rich counterparts. His follow-up work, "From Play to Policy" (2022, 6 citations), extends these ideas to uncurated "play data," enabling robots to learn rich behavioral repertoires without requiring carefully scripted demonstrations. Taken together, Cui's research champions a pragmatic and scalable vision for robot learning: one where messy, naturalistic human data becomes a strength rather than an obstacle, pushing the field closer to deployable, general-purpose robotic agents.