Hongxiang Yu

Zhejiang University of Technology, Zhejiang University

Papers

Total Citations

H-Index

About

Hongxiang Yu is a robotics researcher advancing the frontiers of precision manipulation and autonomous decision-making in real-world environments. His work centers on vision-guided robotic assembly, self-assessment for safe task execution, and neural image servo control—critical areas for deploying robots beyond structured factory floors. Yu’s most influential contribution is his 2022 paper on sub-millimeter peg-in-hole assembly, which introduces a seam-filling strategy inspired by human visual feedback. This work, with 12 citations, enables robots to handle unseen peg shapes with remarkable precision, directly addressing a long-standing challenge in industrial automation. In 2024, Yu proposed the Correspondence Encoded Neural Image Servo Policy (CNS), an architecture that achieves high-precision positioning by learning robust intermediate representations from visual input—a significant step toward closing the gap between classical control and modern learning-based methods. His 2023 work on failure-aware policy learning introduces self-assessment rules that allow robots to verify and re-select actions before execution, enhancing safety in autonomous systems. Together, these contributions establish Yu as a rising leader in vision-based robotic manipulation, with clear impact on both foundational research and practical deployment.