Xianghao Xu

Papers

Total Citations

H-Index

About

Xianghao Xu is a researcher specializing in 3D shape understanding, unsupervised learning, and geometric processing, with a particular focus on advancing how machines perceive and interact with structured three-dimensional objects. His work sits at the intersection of computer vision, robotics, and computer graphics, addressing fundamental challenges in shape representation and articulation. Among his notable contributions, Xu has pioneered unsupervised approaches to complex 3D shape problems. His 2023 work on part retrieval and assembly introduced a method for decomposing 3D shapes into meaningful primitives without requiring labeled training data, enabling downstream applications in robotic manipulation, shape editing, and compression. Complementing this, his 2022 research on kinematic motion detection tackled the labor-intensive challenge of creating articulated 3D object datasets, developing an unsupervised pipeline to automatically identify how object parts move — a capability critical for populating virtual environments and generating synthetic training data for robotics and vision systems. With a growing citation record across both works, Xu's research addresses real bottlenecks in scalable 3D dataset creation and shape understanding. His contributions are particularly valuable for researchers and students working in embodied AI, simulation, and geometric deep learning, offering practical, data-efficient solutions to longstanding challenges in the field.