Xuewei Qi

Papers

Total Citations

H-Index

About

Xuewei Qi is a roboticist advancing the frontier of active perception and object navigation. Her research lies at the intersection of computer vision, reinforcement learning, and motion planning, with a primary focus on enabling robots to intelligently interact with their environments. Qi’s major contribution is the development of "Decision Transformers for Active Object Detection" (2023, 16 citations), a pioneering framework that unifies planning and perception. Unlike traditional systems that treat these as separate modules, her work allows a robot to learn *how* to move in order to improve its own visual understanding, effectively teaching machines to “look” strategically. Building on this, she introduced VLPG-Nav (2024, 2 citations), a novel visual language navigation method that not only guides a robot to a target object but also ensures the object is centered in the camera’s field of view—a critical, often-overlooked challenge for downstream manipulation tasks. By tackling the coupling of motion and vision, Qi is laying the groundwork for more autonomous, perceptive robots capable of operating in complex, unstructured household environments. Her work is essential reading for anyone interested in embodied AI and next-generation robotic perception.