Fanyu Zeng

University of Electronic Science and Technology of China, Nanjing University of Posts and Telecommunications

Papers

Total Citations

113

H-Index

About

Fanyu Zeng is a pioneering researcher at the intersection of embodied AI, deep reinforcement learning (DRL), and autonomous robotics. His primary research focuses on enabling artificial agents to perceive, navigate, and physically interact with unstructured environments, with a special emphasis on visual navigation and robotic manipulation for construction and decoration tasks. Zeng’s most influential work, “A Survey on Visual Navigation for Artificial Agents With Deep Reinforcement Learning” (2020, 96 citations), has become a foundational reference in the field, systematically mapping the emerging paradigm of DRL-driven visual navigation. Beyond surveys, he has made significant technical contributions to low-shot learning and defect detection, developing novel deep learning architectures like WallNet—a hierarchical visual attention model for precise putty bulge terminal point detection. His work on deep transfer learning for wall bulge endpoints regression (2022) and low-shot wall defect detection (2020) directly addresses the practical challenges of deploying autonomous decoration robots in real-world construction sites, where labeled data is scarce. By bridging the gap between simulation and real-world robotic autonomy, Zeng’s research is shaping the next generation of intelligent construction robots capable of performing complex, adaptive tasks with minimal human supervision.