Qingnan Fan
Bellevue Hospital Center, Stanford University, Tencent (China)
Papers
10
Total Citations
173
H-Index
5
About
Qingnan Fan is a researcher working at the intersection of 3D computer vision, robotics, and machine learning, with a particular focus on object manipulation, scene understanding, and autonomous assembly. His work addresses some of the most challenging problems in enabling robots to perceive, reason about, and interact with complex 3D environments. Fan's most influential contribution, "Generative 3D Part Assembly via Dynamic Graph Learning" (2020, 46 citations), introduced a novel framework for intelligent part assembly — essentially teaching AI systems to solve real-world spatial puzzles analogous to assembling furniture. His subsequent research expanded into articulated object manipulation, with works like "VAT-Mart" and "AdaAfford" (36 citations) developing systems that help robots understand how to interact with everyday objects such as doors and cabinets. His investigation of multi-robot coordination, demonstrated in "Multi-Robot Active Mapping via Neural Bipartite Graph Matching" (27 citations), showcases his range across collaborative robotic systems. Additional work on camera relocalization in dynamic environments further highlights his contributions to robust spatial reasoning. With over 170 cumulative citations across a relatively concise body of work, Fan has established himself as a rising voice in embodied AI and robot learning, making meaningful strides toward adaptable, real-world robotic systems.
Research Focus
Key Achievements
Top Papers
- 1Generative 3D Part Assembly via Dynamic Graph Learning46 citations · 2020
- 2
- 3
- 4Multi-Robot Active Mapping via Neural Bipartite Graph Matching27 citations · 2022
- 5
- 6
- 7
- 8
- 9Multi-Robot Active Mapping via Neural Bipartite Graph Matching2 citations · 2022
- 10