Equivariant diffusion policy for sample-efficient robotic manipulation

Dian Wang, Stephen Hart, David Surovik, Tarik Kelestemur, Haojie Huang, Haibo Zhao, Mark Yeatman, Xupeng Zhu, Boce Hu, Mingxi Jia, Jiuguang Wang, Robin Walters, Robert Platt, Northeastern University, Stanford University, Robotics and AI Institute, Brown University

The International Journal of Robotics Research

Published online on February 28, 2026

Abstract

The International Journal of Robotics Research, Ahead of Print.
Recent work has shown diffusion models are an effective approach to learning the multimodal distributions arising from demonstration data in behavior cloning. However, a drawback of this approach is the need to learn a denoising function, which is ...