I-Chun Arthur Liu

Papers

Total Citations

H-Index

About

I-Chun Arthur Liu is a robotics researcher whose work centers on deformable object manipulation, learning from demonstration (LfD), and visuomotor policy learning. His major contribution is the development of DMfD (Deformable Manipulation from Demonstrations), a novel LfD framework that enables robots to handle complex, non-rigid objects like cables, cloth, or ropes using either state or image inputs. This work, published in 2022 and already garnering 32 citations, demonstrates how expert demonstrations can be leveraged in three complementary ways to balance task performance and generalization. Liu’s research addresses a critical gap in robotics—most manipulation systems struggle with deformable objects due to their high-dimensional, underactuated dynamics. By showing that robots can learn effective policies from relatively few demonstrations, his work has practical implications for manufacturing, healthcare, and domestic assistance. His approach stands out for its versatility across input modalities and its ability to transfer skills to novel scenarios. As deformable manipulation remains a frontier in robotics, Liu’s contributions are shaping how researchers think about data-efficient, real-world robot learning.