Chaewon Hwang

Papers

Total Citations

H-Index

About

Chaewon Hwang is a rising researcher at the forefront of generative AI and robotic manipulation, whose work bridges the critical gap between equivariant deep learning and real-world physical interaction. Her primary research areas include SE(3)-equivariant generative modeling, diffusion-based policy learning, and visual robotic manipulation. Hwang’s most significant contribution to date is the development of **Diffusion-EDFs**, a novel framework that integrates bi-equivariant denoising directly on the SE(3) manifold for learning manipulation tasks from stochastic human demonstrations. This work, published in 2024, has already garnered 18 citations, signaling its rapid impact on the field. By ensuring that the diffusion process respects the symmetries of 3D space, Hwang’s method enables robots to generalize manipulation skills—such as grasping and assembly—across diverse object poses without requiring exhaustive retraining. Her approach represents a key step toward more data-efficient and robust robotic learning, addressing a fundamental challenge in embodied AI. As a young scholar, Hwang’s work is poised to influence the next generation of equivariant generative models for robotics, and she is increasingly recognized as a leading voice in the integration of geometric deep learning with physical action.