I am a Ph.D. student in Computer Science at Georgia Tech, advised by Prof. Danfei Xu. Before this, I received my M.S. in Computer Science and B.S. degrees from UIUC, where I was advised by Prof. Kris Hauser and Prof. Yunzhu Li.
My research focuses on Robot Learning.
Specifically, I aim to develop a general-purpose and life-long robotic system that is (1) generalizable across diverse scenarios (e.g. instances, environments, embodiments, and tasks); (2) adaptable to unseen domains efficiently through and for physical interactions; and (3) deployable in the REAL 3D unstructured world.
To this end, I am currently working on:
*: Equal contribution. See also my Google Scholar profile for the most recent publications.
arXiv 2026Selected
Project Paper ArXiv Code (coming) Data Thread
Distinguished Poster Award at Meta Research Summit for Egocentric Intelligence (RSIE), 2026
A systematic study of the World Action Model's state representation for bridging the human-robot embodiment gap, letting policy performance scale with in-the-wild egocentric human data where naive BC co-training fails.
RSS 2026Selected
Project Paper ArXiv Code Data Thread
A full-stack ecosystem for robot learning from egocentric human data: 1,300+ hours of demonstrations, cloud infrastructure, and human-to-robot transfer algorithms, growing with community contributions.
Spotlight at NeurIPS 2025 (top 3.19%)
Language-grounded object orientation that bridges spatial reasoning and manipulation, letting robots follow orientation-aware instructions in open-world scenes.
RSS 2024Selected
Project Paper ArXiv Code Thread
Best Abstract Award at ICRA 2024 RMDO Workshop Link
Learning material-conditioned Graph-based Neural Dynamics (GBND) model with few-shot physical property adaptation for heterogeneous deformable object manipulation.
Meta-learned portfolio warmstarts Bayesian Optimization in AutoMPC, enabling faster, more stable tuning of data-driven model predictive control system.
Noisy positive-unlabeled learning with self-training to robustly perform speculative reasoning over incomplete or noisy knowledge graphs.