Render and Diffuse: Aligning Image and Action Spaces for Diffusion-based Behaviour Cloning

TL;DR

This paper introduces Render and Diffuse (R&D), a novel method that aligns image and action spaces in robot learning using virtual renders, improving sample efficiency and spatial generalization in imitation learning tasks.

Contribution

R&D unifies robot actions and observations in image space via virtual rendering, simplifying learning and enhancing generalization and sample efficiency.

Findings

R&D outperforms traditional image-to-action methods in sample efficiency.

R&D demonstrates strong spatial generalization in real-world tasks.

The method is effective across six diverse robotic tasks.

Abstract

In the field of Robot Learning, the complex mapping between high-dimensional observations such as RGB images and low-level robotic actions, two inherently very different spaces, constitutes a complex learning problem, especially with limited amounts of data. In this work, we introduce Render and Diffuse (R&D) a method that unifies low-level robot actions and RGB observations within the image space using virtual renders of the 3D model of the robot. Using this joint observation-action representation it computes low-level robot actions using a learnt diffusion process that iteratively updates the virtual renders of the robot. This space unification simplifies the learning problem and introduces inductive biases that are crucial for sample efficiency and spatial generalisation. We thoroughly evaluate several variants of R&D in simulation and showcase their applicability on six everyday tasks in the real world. Our results show that R&D exhibits strong spatial generalisation capabilities and is more sample efficient than more common image-to-action methods.