Representation Learning for Robotic Manipulation
I am interested in task-aware representations that preserve the geometric, semantic, and interaction-relevant structure needed for policy learning and robust transfer across objects and tasks.
About
I am a PhD candidate at the Institute of Data Science, The University of Hong Kong (HKU IDS) and an HKU Presidential PhD Scholar (HKUPS). I am advised by Prof. Yanchao Yang and Prof. Yi Ma. Prior to joining HKU, I received my B.S. in Computer Science and Technology from Zhejiang University in 2023.
My research focuses on building better representations for robotic manipulation and on using large language models to construct embodied agents that integrate perception, reasoning, and policy models for robust physical interaction. I am particularly interested in representation learning, vision-language-action models, dexterous manipulation, and agentic robotics.
I am interested in task-aware representations that preserve the geometric, semantic, and interaction-relevant structure needed for policy learning and robust transfer across objects and tasks.
I explore how visual and multimodal observations, together with generative simulation, can support high-dimensional dexterous control and transferable manipulation policies.
I am interested in LLM-based embodied agents that coordinate perception, reasoning, and learned policies to select skills, monitor execution, and adapt during physical interaction.