Equivariant Diffusion Policy

TL;DR

This paper introduces Equivariant Diffusion Policy, a method that incorporates domain symmetries into diffusion models to improve sample efficiency and generalization in behavior cloning, especially for 6-DoF control tasks.

Contribution

It leverages SO(2) symmetry in diffusion models for policy learning, providing theoretical analysis and empirical validation on simulation and real-world tasks.

Findings

21.9% higher success rate than baseline in simulations

Effective policy learning with fewer samples in real-world system

Theoretically characterizes SO(2)-equivariance in diffusion models

Abstract

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 significantly more complex than learning an explicit policy. In this work, we propose Equivariant Diffusion Policy, a novel diffusion policy learning method that leverages domain symmetries to obtain better sample efficiency and generalization in the denoising function. We theoretically analyze the $\mathrm{SO}(2)$ symmetry of full 6-DoF control and characterize when a diffusion model is $\mathrm{SO}(2)$-equivariant. We furthermore evaluate the method empirically on a set of 12 simulation tasks in MimicGen, and show that it obtains a success rate that is, on average, 21.9% higher than the baseline Diffusion Policy. We also evaluate the method on a real-world system to show that effective policies can be learned with relatively few training samples, whereas the baseline Diffusion Policy cannot.