Wang aims to build intelligent multimodal AI agents that can understand the world, collaborate with humans, and perform real-world tasks—from everyday activities to high-stakes missions. His work spans multimodal representation learning, embodied AI for human-agent collaboration, and the ethical design of trustworthy AI systems. Drawing on methodologies from machine learning, computer vision, natural language processing, and robotics—with insights from cognitive science and neuroscience—his research develops generalizable, efficient, and socially responsible agents that perceive, communicate, and act in complex environments.