Zihao Yu 于子皓

Research: Embodied AI, Large Language Models, Robotics


MS in Computer Science | BS in Computer Engineering
BS in Electrical Engineering (Minor in Robotics)
@ Washington University in St. Louis

BA in Physics @ Grinnell College

I am a researcher and engineer working on embodied AI and robotics. I recently completed my M.S. in Computer Science at Washington University in St. Louis, where I developed research on embodied memory, mentored by Prof. Chongjie Zhang. During my undergraduate studies at WashU, I conducted research on language models in the WashU NLP Group, mentored by Prof. Chenguang Wang.

My research focuses on building embodied agents that can perceive, reason, learn from experience, and interact with the physical world. I am particularly interested in robot learning, embodied memory, agentic learning, and foundation models for long-horizon reasoning and decision-making.

My long-term goal is to develop intelligent embodied systems that can continuously learn from their interactions with the world and reliably assist people in real-world environments.

I come from Shijiazhuang, Hebei, China, and came to the U.S. for college. I began my undergraduate studies in 2020 and finished my B.A. in Physics at Grinnell College in 2023, where I was fortunate to be advised by Prof. Keisuke Hasegawa, mentored by Prof. Charles Cunningham, and inspired by every professor I learned from at Grinnell. In Spring 2023, when ChatGPT-3.5 was released, I became fascinated by AI and was accepted into the WashU Dual Degree Program to continue my undergraduate education in Engineering.

At WashU, I studied classical engineering fundamentals while also exploring the frontiers of artificial intelligence. I earned my second and third undergraduate degrees in B.S. in Computer Engineering and B.S. in Electrical Engineering (with a minor in Robotics). I was fortunate to be advised by Prof. Roger Chamberlain and mentored by Prof. Jason Trobaugh. I then spent one more year at WashU to complete the M.S. in Computer Science program.

During my time at WashU, I also had the opportunity to extend my research beyond academia into real-world robotic systems. In the summer of 2025, I joined Tencent Robotics X Lab for my first internship, where I worked on embodied AI and intelligent robotic systems. The experience gave me a new perspective on research—how ideas in embodied intelligence can be translated into systems that operate reliably in complex physical environments and ultimately become useful in people’s everyday lives.

Journey

Research

MEMORA: Embodied Action Memory from Egocentric Videos for Reasoning and Planning
Zihao Yu, Xiu Yuan, Chongjie Zhang
Robotics: Science and Systems (RSS 2026), Workshop on Foundation Models for Robot Planning (FM4RoboPlan) — Oral Presentation
Empirical Methods in Natural Language Processing (EMNLP 2026)
TAIROS: An Embodied AI Platform for Robotics Applications
Tencent Robotics X Team & Futian Laboratory
2025
From Language to Action: A Vision-Language and Kinematics-Guided Framework for 6-DOF Robotic Arm Manipulation
Zihao Yu
Undergraduate Capstone Thesis, Department of Electrical & Systems Engineering, Washington University in St. Louis, Spring 2025
COSMIC: Generalized Refusal Direction Identification in LLM Activations
Vincent Siu, Nicholas Crispino, Zihao Yu, Sam Pan, Zhun Wang, Yang Liu, Dawn Song, Chenguang Wang
The Annual Meeting of the Association for Computational Linguistics (ACL 2025)
MLAN: Language-Based Instruction Tuning Preserves and Transfers Knowledge in Multimodal Large Language Models
Jianhong Tu*, Zhuohao Ni*, Nicholas Crispino, Zihao Yu, Michael Bendersky, Beliz Gunel, Ruoxi Jia, Xin Liu, Lingjuan Lyu, Dawn Song, Chenguang Wang
KnowFM @ ACL 2025