Tung-I Chen

Papers

Total Citations

H-Index

About

Tung-I Chen is a robotics and computer vision researcher whose work centers on solving fundamental manipulation and perception challenges. His primary research areas include robotic assembly, visual servoing, and 3D point cloud registration, with a particular focus on achieving high-precision, object-agnostic manipulation in unstructured environments. Chen’s major contributions lie in developing coarse-to-fine frameworks that bridge the gap between global reasoning and local precision. His work on CFVS (Coarse-to-Fine Visual Servoing) addresses the notoriously difficult peg-in-hole assembly problem by enabling 6-DoF manipulation without object-specific models, overcoming limitations of prior approaches that restricted degrees of freedom or required close initial positioning. This work has garnered 14 citations since 2023, reflecting its immediate impact on the robotics community. In parallel, Chen has advanced point cloud registration through his SE(3)-equivariant representation learning approach, which jointly reasons about global shapes and local geometric features to achieve robust alignment even under partial overlap and significant pose differences. This paper has accumulated 13 citations, demonstrating its relevance to both computer vision and robotics applications. Chen’s work represents a significant step toward more generalizable robotic manipulation systems that can operate reliably in real-world, unconstrained settings.