Hanyang Chen

My research develops AI-driven methods for urban traffic management, with a focus on traffic signal control. I build reinforcement learning approaches that address real-world challenges such as missing traffic data and proactive decision-making, aiming to improve traffic flow and reduce congestion in dynamic urban environments.

I study how to train reinforcement learning agents that can adapt to target environments whose dynamics differ from those seen during training. The goal is to improve policy robustness and generalization when deploying RL systems in new domains with varying dynamics.

I investigate how Vision-Language-Action (VLA) models integrate visual perception, language understanding, and action generation for embodied agents. My focus is on improving their semantic understanding of visual and linguistic instructions so that agents can make more reliable decisions in interactive environments.