Siyan Dong

Papers

Total Citations

H-Index

About

Siyan Dong is a rising researcher in computer vision and robotics, with a focus on active perception, multi-agent coordination, and visual localization. Their work tackles fundamental challenges in enabling autonomous systems to efficiently understand and navigate complex environments. In a key contribution, Dong introduced a neural bipartite graph matching approach for multi-robot active mapping, which optimizes goal position estimation to minimize the time required for complete scene reconstruction—a critical advancement for search-and-rescue or exploration missions. This work, published in 2022, has already garnered early citations for its novel formulation. Dong has also advanced visual localization through a few-shot scene region classification method, enabling robust 6-DoF camera pose estimation with minimal training data, a vital capability for augmented reality and autonomous navigation. By leveraging memory-based techniques, this approach reduces reliance on large-scale 3D maps. With these contributions, Dong is establishing a reputation for integrating learning-based methods with geometric reasoning, addressing real-world constraints like data efficiency and multi-robot coordination. Their work promises to shape the next generation of intelligent, autonomous systems.