Xin Zhang

CityLens: Evaluating Large Vision-Language Models for Urban Socioeconomic Sensing

MoWM: Mixture-of-World-Models for Embodied Planning via Latent-to-Pixel Feature Modulation

LongScape: Advancing Long-Horizon Embodied World Models with Context-Aware MoE

RoboScape: Physics-informed Embodied World Model

CityBench: Evaluating the Capabilities of Large Language Models for Urban Tasks

UrbanMLLM: Joint Learning of Cross-view Imagery for Urban Understanding