VisuoSpatial Foresight for Multi-Step, Multi-Task Fabric Manipulation

TL;DR

This paper introduces VisuoSpatial Foresight (VSF), a simulation-trained visual dynamics model that generalizes fabric manipulation tasks, significantly improving success rates by leveraging RGBD data in multi-step, multi-task scenarios.

Contribution

The paper presents VSF, a novel framework that learns fabric dynamics from RGB and depth images in simulation, enabling generalization across multiple fabric manipulation tasks without demonstrations.

Findings

RGBD data improves folding success by 80% over RGB alone.

VSF outperforms 5 baseline methods in simulation and real robot tests.

Depth information significantly enhances fabric manipulation performance.

Abstract

Robotic fabric manipulation has applications in home robotics, textiles, senior care and surgery. Existing fabric manipulation techniques, however, are designed for specific tasks, making it difficult to generalize across different but related tasks. We extend the Visual Foresight framework to learn fabric dynamics that can be efficiently reused to accomplish different fabric manipulation tasks with a single goal-conditioned policy. We introduce VisuoSpatial Foresight (VSF), which builds on prior work by learning visual dynamics on domain randomized RGB images and depth maps simultaneously and completely in simulation. We experimentally evaluate VSF on multi-step fabric smoothing and folding tasks against 5 baseline methods in simulation and on the da Vinci Research Kit (dVRK) surgical robot without any demonstrations at train or test time. Furthermore, we find that leveraging depth significantly improves performance. RGBD data yields an 80% improvement in fabric folding success rate over pure RGB data. Code, data, videos, and supplementary material are available at https://sites.google.com/view/fabric-vsf/.