PACA: Perspective-Aware Cross-Attention Representation for Zero-Shot Scene Rearrangement

TL;DR

PACA is a zero-shot scene rearrangement method that uses perspective-aware cross-attention from Stable Diffusion to generate object representations and control viewpoints, enabling more accurate robot task execution.

Contribution

It introduces a unified, perspective-aware cross-attention representation for scene rearrangement that integrates generation, segmentation, and encoding, extending to 6-DoF views in a zero-shot setting.

Findings

Achieves 87% average matching accuracy in real robot experiments.

Attains 67% execution success rate across various scenes.

Supports 6-DoF camera view matching, surpassing previous 3-DoF limitations.

Abstract

Scene rearrangement, like table tidying, is a challenging task in robotic manipulation due to the complexity of predicting diverse object arrangements. Web-scale trained generative models such as Stable Diffusion can aid by generating natural scenes as goals. To facilitate robot execution, object-level representations must be extracted to match the real scenes with the generated goals and to calculate object pose transformations. Current methods typically use a multi-step design that involves separate models for generation, segmentation, and feature encoding, which can lead to a low success rate due to error accumulation. Furthermore, they lack control over the viewing perspectives of the generated goals, restricting the tasks to 3-DoF settings. In this paper, we propose PACA, a zero-shot pipeline for scene rearrangement that leverages perspective-aware cross-attention representation derived from Stable Diffusion. Specifically, we develop a representation that integrates generation, segmentation, and feature encoding into a single step to produce object-level representations. Additionally, we introduce perspective control, thus enabling the matching of 6-DoF camera views and extending past approaches that were limited to 3-DoF top-down views. The efficacy of our method is demonstrated through its zero-shot performance in real robot experiments across various scenes, achieving an average matching accuracy and execution success rate of 87% and 67%, respectively.