Zunzhe Zhang

I am a B.S. candidate in Computer Science at the Institute for Interdisciplinary Information Sciences, Tsinghua University. My research focuses on embodied intelligence: agents that connect perception, language, memory, and continuous control to act reliably in physical or simulated environments.

Recently, I have been working on Vision-Language-Action foundation models, flow-matching control policies, LLM-based embodied agents, and interactive multi-agent simulation. I am also interested in evaluation and simulation tooling: how to expose meaningful failures, generate useful data, and make robotic policies more robust outside narrow demos.

I previously worked on embodied AI research with Prof. Chuang Gan at UMass Amherst, contributing to projects on LLM agents and virtual communities. I also worked as a VLA algorithm intern at Galaxea Inc., where I explored continuous action generation and adaptive robotic control.