William Shen

Papers

Total Citations

H-Index

About

William Shen is a leading researcher in embodied AI and robotic manipulation, with a focus on enabling robots to navigate and interact intelligently in cluttered, real-world environments. His most impactful work includes the **Interactive Gibson Benchmark**, the first comprehensive benchmark for training and evaluating **Interactive Navigation**—a paradigm where robots are encouraged to physically move objects to accomplish tasks, rather than avoiding them. This benchmark, with over 15 combined citations, has become a foundational tool for the field, pushing robots beyond passive navigation toward active, problem-solving behavior. More recently, Shen introduced **Distilled Feature Fields for Few-Shot Language-Guided Manipulation**, a breakthrough that bridges the gap between 2D vision-language models and 3D robotic control. By distilling rich semantic features from self-supervised models into 3D representations, his method allows robots to generalize manipulation skills from just a few demonstrations, guided by natural language commands. This work, already garnering attention, promises to make robotic systems far more adaptable and intuitive. Shen’s contributions are shaping the next generation of robots that can understand, navigate, and manipulate the messy, human-centric world.