Xingyu Lin
Papers
6
Total Citations
91
H-Index
4
About
Xingyu Lin is a robotics researcher whose work spans robotic manipulation, deformable object handling, and learning-based control systems. His research addresses some of the most challenging perception and manipulation problems in robotics, with particular emphasis on cloth manipulation, liquid state estimation, and cutting of complex multi-material objects. Lin's most-cited contribution, "Mesh-based Dynamics with Occlusion Reasoning for Cloth Manipulation" (2022, 37 citations), tackles the difficult problem of self-occlusion in cloth unfolding by integrating advances in pose estimation with dynamic modeling. His work on transparent liquid segmentation (2022, 20 citations) introduced a self-supervised pipeline requiring no manual annotation — a significant practical advance for robotic pouring tasks. His RoboNinja system (2023, 16 citations) demonstrated adaptive, closed-loop cutting policies for multi-material objects, moving beyond simplistic open-loop approaches. Through DiffSkill (2022, 14 citations), he leveraged differentiable physics simulators to enable efficient tool-based manipulation of deformable objects. More recently, Lin has expanded into teleoperation frameworks and humanoid robot benchmarking, reflecting a broadening research agenda. Collectively, his work demonstrates a consistent drive to bring robust perception and adaptive learning to unstructured, real-world robotic manipulation challenges.
Research Focus
Key Achievements
Top Papers
- 1Mesh-based Dynamics with Occlusion Reasoning for Cloth Manipulation37 citations · 2022
- 2Self-supervised Transparent Liquid Segmentation for Robotic Pouring20 citations · 2022
- 3RoboNinja: Learning an Adaptive Cutting Policy for Multi-Material Objects16 citations · 2023
- 4
- 5
- 6