Delong Zhu
Papers
24
Total Citations
848
H-Index
12
About
Delong Zhu is a robotics and computer vision researcher whose work spans autonomous robot exploration, visual SLAM (Simultaneous Localization and Mapping), and deep reinforcement learning for intelligent navigation. He is perhaps best known as a contributor to **TartanAir** (2020), a landmark simulation dataset designed to stress-test visual SLAM systems under challenging conditions including dynamic objects, varying lighting, and diverse weather — a work that has amassed over 365 citations and become a key benchmark in the field. Zhu has made significant strides in autonomous exploration, developing deep reinforcement learning frameworks that improve long-term planning over greedy approaches (108 citations) and semantic road map strategies for efficient indoor navigation (72 citations). His **HouseExpo** dataset (2020, 62 citations) further demonstrates his commitment to building community resources for mobile robotics research. Beyond navigation, Zhu has tackled practical service robotics challenges, including elevator button recognition using convolutional neural networks, and has contributed to medical robotics through novel registration techniques for computer-assisted orthopedic surgery. His body of work reflects a researcher dedicated to bridging simulation, perception, and real-world robotic autonomy.
Research Focus
Key Achievements
Top Papers
- 1TartanAir: A Dataset to Push the Limits of Visual SLAM365 citations · 2020
- 2
- 3
- 4
- 5An Object SLAM Framework for Association, Mapping, and High-Level Tasks47 citations · 2023
- 6A Novel OCR-RCNN for Elevator Button Recognition30 citations · 2018
- 7TartanAir: A Dataset to Push the Limits of Visual SLAM26 citations · 2020
- 8
- 9
- 10