Soomi Lee
Papers
1
Total Citations
6
H-Index
1
About
Soomi Lee is a leading researcher at the intersection of geometric deep learning and robotics, with a primary focus on SE(3)-equivariant representations for robot learning and control. Her seminal work, the tutorial survey "SE(3)-equivariant Robot Learning and Control," published in 2025, has already garnered 6 citations, establishing a foundational framework for incorporating symmetry and group theory into robotic manipulation and motion planning. Lee’s major contribution lies in demonstrating how equivariant neural networks can dramatically improve sample efficiency, generalization, and robustness in tasks ranging from grasping to dexterous manipulation, by exploiting the inherent rotational and translational symmetries of physical systems. Her research bridges theoretical advances in geometric deep learning with practical robotic applications, offering a principled approach to learning policies that are inherently invariant to viewpoint and object pose. Lee’s work is particularly notable for its clarity and accessibility, making complex mathematical concepts actionable for the robotics community. As a rising star in the field, she is shaping how robots perceive and interact with the world, with her tutorial serving as a key resource for students and researchers seeking to integrate equivariance into their own work.
Research Focus
Key Achievements
Top Papers
- 1SE(3)-equivariant Robot Learning and Control: A Tutorial Survey6 citations · 2025