SpeedFolding: Learning Efficient Bimanual Folding of Garments

TL;DR

SpeedFolding introduces a neural network-based bimanual system that efficiently folds garments from crumpled states to folded configurations, significantly outperforming prior methods in speed and success rate.

Contribution

The paper presents a novel neural network architecture for predicting bimanual actions, enabling rapid and reliable garment folding from diverse initial states.

Findings

Achieves 93% success rate in garment folding

Folds garments in under 120 seconds on average

Reaches 30-40 Folds Per Hour, outperforming prior work

Abstract

Folding garments reliably and efficiently is a long standing challenge in robotic manipulation due to the complex dynamics and high dimensional configuration space of garments. An intuitive approach is to initially manipulate the garment to a canonical smooth configuration before folding. In this work, we develop SpeedFolding, a reliable and efficient bimanual system, which given user-defined instructions as folding lines, manipulates an initially crumpled garment to (1) a smoothed and (2) a folded configuration. Our primary contribution is a novel neural network architecture that is able to predict pairs of gripper poses to parameterize a diverse set of bimanual action primitives. After learning from 4300 human-annotated and self-supervised actions, the robot is able to fold garments from a random initial configuration in under 120s on average with a success rate of 93%. Real-world experiments show that the system is able to generalize to unseen garments of different color, shape, and stiffness. While prior work achieved 3-6 Folds Per Hour (FPH), SpeedFolding achieves 30-40 FPH.