Xuheng Zhang

I'm Xuheng Zhang, at EECS in Peking University, currently a sophomore student in the Intelligent Science and Technology program (Zhi Class) at Peking University.

My long-term goal is to build robotic systems that generalize broadly in complex, human-centered environments. I believe scalable intelligence requires moving beyond imitation to acquire fundamental world dynamics during pre-training—the key to transferable physical understanding. Currently, I focus on pre-training and RL post-training of VLA and world model-based foundation models to embed these dynamics for robust manipulation, ultimately aiming to unify high-level perception, planning, and low-level action within a physically grounded framework.

Apart from my research interests, I enjoy rock climbing, I'm a v3 level climber, and I find it a great way to challenge myself both physically and mentally. Besides, I also enjoy traveling, photography and ultimate frisbee.

Email Github Google Scholar

2024-2025: National Scholarship (Highest honor for undergraduates in China)

LDA-1B: Scaling Latent Dynamics Action Model via Universal Embodied Data Ingestion

Jiangran Lyu*, Kai Liu*, Xuheng Zhang*, Haoran Liao, Yusen Feng, Wenxuan Zhu, Tingrui Shen, Jiayi Chen, Jiazhao Zhang, Yifei Dong, Wenbo Cui, Senmao Qi, Shuo Wang, Yixin Zheng, Mi Yan, Xuesong Shi, Haoran Li, Dongbin Zhao, Ming-Yu Liu, Zhizheng Zhang^†, Li Yi^†, Yizhou Wang^†, He Wang^†

[Arxiv] [Code] [Project page]

GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Data

Shengliang Deng*, Mi Yan*, Songlin Wei, Haixin Ma, Yuxin Yang, Jiayi Chen, Zhiqi Zhang, Taoyu Yang, Xuheng Zhang, Heming Cui, Zhizheng Zhang^†, He Wang^†

[Arxiv] [Code] [Project page]

GALBOT 2025.05 - Present

Research Intern

Research Advisor: Zhizheng Zhang, He Wang

Peking University (PKU) 2024.09 - 2028.07 (expected)

EECS

Zhi Class