Tsung-Yi Lin
Papers
2
Total Citations
150
H-Index
2
About
Tsung-Yi Lin is a leading researcher in computer vision and robotics, best known for bridging the gap between visual perception and robotic manipulation. His work centers on developing representations that allow robots to understand and interact with complex, real-world objects. A key contribution is the introduction of NeRF-Supervision, a method that leverages neural radiance fields to learn dense object descriptors from RGB images alone. This breakthrough is particularly impactful for robot perception, as it enables robust handling of notoriously difficult objects—such as thin, reflective, or transparent items like forks and whisks—which often defeat traditional RGB-D or multi-view stereo pipelines. Lin’s research has also explored the fundamental question of how visual priors influence manipulation learning. In his highly cited work “Learning to See before Learning to Act,” he demonstrated that pre-training on passive vision tasks can significantly accelerate the acquisition of manipulation skills, providing a powerful framework for transfer learning in robotics. With over 150 citations across his top papers, Lin’s contributions are shaping how robots perceive and interact with their environment, moving beyond idealized lab settings toward practical, everyday applications.
Research Focus
Key Achievements
Top Papers
- 1
- 2Learning to See before Learning to Act: Visual Pre-training for Manipulation66 citations · 2020