Gehao Zhang

I am an M.S. student in Electrical and Computer Engineering at Northwestern University, advised by Qi Zhu, where I work on embodied AI and 3D vision. I am currently a visiting researcher at the University of Washington, working with Jiafei Duan and Ranjay Krishna.

I received my B.S. in Robot Engineering from Southeast University.

Email (gehaozhang2027 [AT] u.northwestern.edu) Google Scholar Twitter/X GitHub LinkedIn

𝕏 Follow @gehao_zhang_

Research Highlights

Grounded Action Model: 3D Grounding as a Foundation for Robotics [Sep 2026]

My research focuses on embodied AI and 3D vision, with the aim of finding general representations for robots. Towards this goal, I am currently interested in:

Foundation models for embodied AI. What a robot policy should inherit from pretrained models, and how far a frozen backbone can carry it.
3D vision for robotics. Object-centric, metric scene understanding that tells a policy which objects matter and where they are.

News

Sep 2026Acknowledged by Understanding AI’s Kai Williams for shaping his piece “OpenAI’s Astra model is shockingly good at robotics”.
Sep 2026Grounded Action Model is on arXiv, with a project page and real-robot videos.
Jun 2026EmboAlign is accepted to IROS 2026.
Jun 2026Joined the University of Washington, Seattle as a visiting researcher.
May 2026EmboAlign is accepted as a Poster at the MMRAgI Workshop at CVPR 2026.
Sep 2025Started the M.S. program in ECE at Northwestern University.
Jun 2025Graduated from Southeast University with a B.S. in Robot Engineering.
Sep 2024RADA is accepted to ICONIP 2024 for an Oral Presentation.
Jun 2023Silver medal (15th / 494 teams) in the Kaggle Image Matching Challenge 2023, CVPR workshop.

Research

Grounded Action Model: 3D Grounding as a Foundation for Robotics

Gehao Zhang, Weikai Huang, Shailesh Shailesh, Yiyan Peng, Jiafei Duan^†, Ranjay Krishna^† (†Equal advising)

arXiv preprint, 2026

Project Page / Paper / Code / Twitter/X

EmboAlign pipeline: compositional constraints guide video selection and trajectory optimization

EmboAlign: Aligning Video Generation with Compositional Constraints for Zero-Shot Manipulation

Gehao Zhang^*, Zhenyang Ni^*, Payal Mohapatra, Han Liu, Ruohan Zhang, Qi Zhu (*Equal contribution)

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2026

Poster at the MMRAgI Workshop at CVPR 2026

Paper

RADA feature matching compared with D2-Net, ASLFeat and SuperPoint

RADA: Robust and Accurate Feature Learning with Domain Adaptation

Jingtai He, Gehao Zhang, Tingting Liu, Songlin Du

International Conference on Neural Information Processing (ICONIP), 2024

Oral Presentation

Paper

Selected Honors

Silver Medal, Kaggle Image Matching Challenge 2023 (CVPR workshop), 15th of 494 teams
“THE BEST” Student Scholarship, Southeast University (top 12%)

Experience

University of Washington, Seattle
Visiting ResearcherJun 2026 – present
Northwestern University
M.S. in Electrical and Computer EngineeringSep 2025 – present
University of Cambridge
Summer sessionAug 2024
University of California, Los Angeles
Summer sessionJun – Aug 2023
Southeast University
B.S. in Robot EngineeringSep 2021 – Jun 2025

Teaching and Service

Teaching assistant, freshman C++ summer course, Southeast University (summer 2023, summer 2024)
President, LabVIEW Club, Southeast University (2023 – 2024)