Research

My research focuses on 3D computer vision and world models, with the long-term goal of building spatial intelligence grounded in the real world. I study how to reconstruct and understand 3D scenes under challenging conditions, including sparse observations and ambiguities from repeated or symmetric structures. Building on this, I am interested in world models that reason over space, time, and physical plausibility, enabling models to infer missing 3D structure and predict future scene evolution beyond what is directly visible. More broadly, I aim to bridge 3D spatial understanding with the semantic reasoning capabilities of vision-language models, toward AI systems that can understand and reason about complex real-world scenes.