Jiwoo Kim

Papers

Total Citations

H-Index

About

Jiwoo Kim is a rising researcher at the forefront of robotic manipulation and generative modeling, with a focus on integrating geometric deep learning with diffusion-based methods. Their key research areas include SE(3)-equivariant neural networks, denoising diffusion probabilistic models, and visuomotor policy learning for robotics. Kim’s major contribution is the development of Diffusion-EDFs, a novel framework that introduces bi-equivariant denoising generative modeling on the SE(3) group for visual robotic manipulation. This work addresses the critical challenge of learning manipulation tasks from stochastic human demonstrations while preserving rotational and translational equivariance, enabling more sample-efficient and generalizable policy learning. Although early in their career, Kim’s work has already garnered 18 citations for this 2024 paper, signaling strong interest from the robotics and machine learning communities. Their approach bridges the gap between equivariant representation learning and diffusion-based policy generation, offering a principled pathway for robots to acquire complex manipulation skills from limited demonstrations. Kim’s contributions are particularly notable for advancing the theoretical foundations of equivariant generative models in robotics, positioning them as an emerging leader in this interdisciplinary field.