Zhile Ren

Georgia Institute of Technology, Brown University

Papers

Total Citations

H-Index

About

Zhile Ren is a leading researcher in embodied AI and robotic perception, whose work bridges the gap between egocentric sensing and allocentric world understanding. His primary contributions lie in semantic mapping and robust simultaneous localization and mapping (SLAM) for dynamic environments. Ren is best known for developing **Semantic MapNet**, a pioneering framework that enables an embodied agent—whether a robot or an AI assistant—to construct a top-down semantic map of an unfamiliar environment from egocentric RGB-D camera observations. This work, which has garnered over 70 combined citations, addresses the fundamental challenge of “what is where?” in spatial AI, allowing agents to build persistent, allocentric representations from a first-person tour. Earlier in his career, Ren tackled the critical problem of navigation in dynamic settings with his work on robust graph SLAM, introducing methods to handle moving landmarks—a key advancement for human-robot interaction. His research has been recognized for its practical impact on autonomous navigation and embodied intelligence, making him a notable figure in the field. Ren’s work continues to inspire new approaches to how machines perceive, map, and interact with complex, changing environments.