William Yang Wang
Papers
5
Total Citations
535
H-Index
4
About
William Yang Wang is a leading researcher at the intersection of natural language processing, computer vision, and robotics, with a core focus on embodied AI and grounded language understanding. His most impactful work tackles the grand challenge of enabling robots to understand and execute complex human instructions in real-world environments. Wang pioneered the REVERIE benchmark (297 citations), a seminal dataset that pushes the frontier of remote embodied visual referring expression, requiring agents to navigate unseen indoor spaces and identify objects based solely on natural language descriptions. He also introduced the influential "Look Before You Leap" framework (200+ citations), which bridges model-free and model-based reinforcement learning for vision-and-language navigation, allowing agents to plan ahead rather than react impulsively. This work directly addresses the critical limitations of purely model-free approaches in dynamic, real-world settings. Beyond English-centric tasks, Wang has advanced cross-lingual vision-language navigation, broadening the accessibility of human-robot interaction. His contributions are foundational to the growing field of embodied AI, consistently shaping how researchers design agents that can perceive, reason, and act in the physical world through language.
Research Focus
Key Achievements
Top Papers
- 1REVERIE: Remote Embodied Visual Referring Expression in Real Indoor Environments297 citations · 2020
- 2
- 3
- 4
- 5Cross-Lingual Vision-Language Navigation2 citations · 2019