Xiwu Zhang

Nanjing University of Science and Technology, University of Wollongong

Papers

Total Citations

370

H-Index

About

Xiwu Zhang is a leading researcher in robotics and computer vision, with a primary focus on visual place recognition and simultaneous localization and mapping (SLAM). His work bridges the gap between traditional geometric methods and modern deep learning, fundamentally advancing how autonomous systems perceive and navigate their environments. Zhang’s most influential contribution is his comprehensive survey, "Visual place recognition: A survey from deep learning perspective" (2020), which has garnered over 250 citations and serves as a foundational reference for researchers entering the field. He is also recognized for pioneering deep learning-based loop closure detection in visual SLAM systems, as demonstrated in his 2017 paper (68 citations), where he replaced hand-crafted features and bag-of-visual-words with convolutional neural networks to achieve more robust and accurate place recognition. Further extending this work, his 2018 study on graph-based place recognition using CNN features (52 citations) introduced a novel framework for leveraging image sequences, enhancing spatial consistency and scalability. Zhang’s research has directly influenced the development of reliable long-term autonomy in robots and autonomous vehicles, making him a key figure in the transition from classical to learning-based visual navigation.