VGGT-World: Transforming VGGT into an Autoregressive Geometry World Model
arXiv · 2026 · Scholar
2nd-year Ph.D. Candidate, School of Electrical Engineering and Computer Science
The University of Queensland
My research focuses on 4D world modeling, aiming to bridge faithful reconstruction and generative inference within a unified framework. My long-term goal is to build general-purpose world models that enable intelligent agents to not only perceive the world as it is, but also reason about what is unseen and anticipate how it evolves.
This vision aligns with emerging world-model efforts such as Genie 3 and PointWorld, which demonstrate the potential of world models to simulate interactive environments for scalable agent learning. My work lies at the intersection of vision foundation models, world models, and embodied AI, with applications in autonomous systems and robotics.
Ordered by year (newest first). Links point to arXiv, DOI, or Google Scholar as available.
arXiv · 2026 · Scholar
CVPR · 2026 · Scholar
CVPR · 2026 · Scholar
arXiv · 2025 · Scholar
NeurIPS · 2025 · Scholar
AAAI · 2025 · Scholar
arXiv · 2024 · Scholar
TCSVT · 2024 · Scholar
T-IV · 2024 · Scholar
ICME · 2023 · Oral · Scholar
ACM MM · 2022 · Scholar
Ph.D. Candidate, Computer Science · Brisbane, Australia
Advisors: Dr. Yadan Luo · Prof. Helen Huang
M.E. (Research), School of Software Engineering · Shanghai, China
GPA 91.02/100 · Thesis: Research on Key Techniques for 3D Reconstruction of Indoor Parking Scenes
Core: Pattern Recognition 100/100; Academic English Writing 94/100
B.E., School of Software · Jinan, China
GPA 4.55/5.0 (top 4%)
Probability and Statistics 100/100; Linear Algebra 95/100; Discrete Mathematics (bilingual) 95/100; Data Structures (bilingual) 95/100; Machine Learning 95/100; Optimization Methods 95/100; Operating System (bilingual) 97/100; Object-oriented Development 97/100