Hong-in Lee
Papers
1
Total Citations
11
H-Index
1
About
Hong-in Lee is a rising researcher in robotics and machine learning, whose work focuses on imbuing robotic systems with geometric awareness to dramatically improve learning efficiency. His primary research areas include SE(3)-equivariant deep learning, visual robotic manipulation, and energy-based models for end-to-end control. Lee’s most notable contribution is the development of Equivariant Descriptor Fields, a framework that exploits spatial roto-translation equivariance to enable sample-efficient learning of complex manipulation tasks from visual input. By embedding SE(3) symmetry directly into neural network architectures, his approach reduces the number of demonstrations required for training, addressing a critical bottleneck in end-to-end robotic learning. This work, published in 2022, has already garnered 11 citations, signaling its early impact on the field. Lee’s research bridges theoretical geometric deep learning with practical robotics, offering a principled path toward more data-efficient and generalizable manipulation policies. His achievements position him as a key contributor to the growing intersection of equivariant neural networks and embodied AI, with potential to influence future robot learning systems.
Research Focus
Key Achievements
Top Papers
- 1