Enabling Stateful Behaviors for Diffusion-based Policy Learning

TL;DR

This paper introduces Diff-Control, a diffusion-based policy that incorporates statefulness via a Bayesian approach, significantly improving robustness and success rates in robot policy learning tasks.

Contribution

It presents a novel stateful diffusion-based policy using Bayesian formulation and ControlNet, enhancing consistency and robustness in robot policy learning.

Findings

Achieves 72% success rate on stateful tasks.

Achieves 84% success rate on dynamic tasks.

Demonstrates improved robustness over existing methods.

Abstract

While imitation learning provides a simple and effective framework for policy learning, acquiring consistent actions during robot execution remains a challenging task. Existing approaches primarily focus on either modifying the action representation at data curation stage or altering the model itself, both of which do not fully address the scalability of consistent action generation. To overcome this limitation, we introduce the Diff-Control policy, which utilizes a diffusion-based model to learn the action representation from a state-space modeling viewpoint. We demonstrate that we can reduce diffusion-based policies' uncertainty by making it stateful through a Bayesian formulation facilitated by ControlNet, leading to improved robustness and success rates. Our experimental results demonstrate the significance of incorporating action statefulness in policy learning, where Diff-Control shows improved performance across various tasks. Specifically, Diff-Control achieves an average success rate of 72% and 84% on stateful and dynamic tasks, respectively. Project page: https://github.com/ir-lab/Diff-Control