Zhizhen Zhang
I am a PhD candidate at The University of Queensland, supervised by Dr. Yadan Luo and Prof Helen Huang. I received my master's degree from Tsinghua University.
I work on embodied AI and robot learning, particularly data-efficient adaptation and model merging for vision-language-action policies. I am interested in how robots can share skills across tasks and adapt to new environments.
News
Earlier news
Selected research
PolicyWeave
arXiv · 2026Unifying independently trained VLA experts through a shared action interface.
MobiGuide
arXiv · 2026Vocabulary-guided trajectory refinement and online recovery for mobile manipulation.
MergeVLA
CVPR · 2026Combining skills from specialist vision-language-action models through model merging.
AcTOL
NeurIPS · 2025Learning temporally ordered and continuous representations from human action videos for robot learning.
Academic service
2027Reviewer · ICLR 2027
2026Reviewer · NeurIPS 2026, CVPR 2026
2025Reviewer · NeurIPS 2025
Teaching
Sem1, 2026INFS4205/7205: Advanced Techniques for High Dimensional Data, EECS, University of Queensland.
Awards & support
2026CVPR Compute Transparency Champion Award · MergeVLA.
2026CVPR Student Travel Grant · MergeVLA.
2025NeurIPS Financial Assistance Grant · AcTOL.
2025Distinguished Poster Award on UQ DS EOY.
2025UQ Research Training Stipend Earmarked.
Academic activities
20253MT presentation at Australasian Database Conference (ADC).
2025Showcase at UQ EECS Illuminate.
2025Co-supervisor for UQ Winter Research Program on the project "Deploying Learning-Based Locomotion on Unitree Go2 for Embodied AI."
2025Presentation at UQ AI Conference Dry-run Seminar.