Research Analyzer
← Back ICRA 2026

SIS: Seam-Informed Strategy for T-Shirt Unfolding

Xuzhao Huang, Akira Seino, Fuyuki Tokuda, Akinari Kobayashi, Dayuan Chen, Yasuhisa Hirata, Norman Tien, Kazuhiro Kosuge

PDF

AI summary

Key figure (auto-extracted from paper)
Leveraging garment seam features significantly improves robotic T-shirt unfolding efficiency and generalization without relying on simulation.
Garment unfolding Robotic manipulation Seam detection Grasping point selection Real-world robotics Decision matrix

Problem

Existing robotic garment unfolding methods struggle with grasping point selection, often relying on simulated data or ignoring structural features like seams, which limits real-world performance and efficiency.

Approach

The proposed Seam-Informed Strategy (SIS) detects seam segments and crossings from RGB images and uses an iterative decision matrix to select optimal grasping points based on human demonstrations and real robot feedback.

Key results

  • Accurate extraction of oriented seam segments and crossings via YOLOv3
  • Decision matrix scoring bridges human demonstrations and real robot execution
  • High unfolding coverage achieved with fewer robot action steps
  • Strong generalization across T-shirt configurations without simulation training

Why it matters

This approach provides a practical, data-driven framework for automating soft material manipulation, benefiting robotics researchers and manufacturers working on garment handling and automation.

Abstract

Seams are information-rich components of garments. The presence of different types of seams and their combinations helps to select grasping points for garment handling. In this let- ter, we propose a new Seam-Informed Strategy (SIS) for finding actions for handling a garment, such as grasping and unfolding a T-shirt. Candidates for a pair of grasping points for a dual-arm manipulator system are extracted using the proposed Seam Feature Extraction Method (SFEM). A pair of grasping points for the robot system is selected by the proposed Decision Matrix Iteration Method (DMIM). The decision matrix is first computed by mul- tiple human demonstrations and updated by the robot execution results to improve the grasping and unfolding performance of the robot. Note that the proposed scheme is trained on real data with- out relying on simulation. Experimental results demonstrate the effectiveness and generalization ability of the proposed strategy.

Index terms

Perception for Grasping and Manipulation Bimanual Manipulation Manipulation Planning

Related papers