Lu Xia

Amazon (United States), The University of Texas at Austin

Papers

Total Citations

170

H-Index

About

Lu Xia is a researcher whose work bridges human-robot interaction, computer vision, and scene understanding, with a particular focus on enabling robots to perceive and interpret human behavior in real-world environments. Her most recognized contributions center on robot-centric activity recognition from first-person perspectives, a technically challenging domain that equips mobile robots with the social and contextual awareness needed to interact naturally with people. Her 2015 paper on activity prediction from first-person videos, which has garnered over 100 citations, introduced a pioneering methodology for early recognition of human activities from a robot's own viewpoint — a critical capability for anticipatory human-robot collaboration. A companion study utilizing RGB-D data further extended this framework by incorporating depth information, accumulating over 50 citations and demonstrating the value of multimodal sensing in social robotics. More recently, Xia has expanded her research into neural scene representation, contributing to the emerging field of Multimodal Neural Radiance Fields (NeRF) for robot vision and 3D scene understanding. Together, her body of work reflects a sustained commitment to advancing robotic perception, making meaningful strides toward robots that can comprehend and respond intelligently to human-centered environments.