A Garment Handling System

Robotic garment handling is needed in various fields but remains a challenging task due to the complex nature of garment states and the inherent deformability of fabrics. The dynamic and unpredictable behavior of textiles requires sophisticated solutions that can adapt to changing conditions and ensure precise manipulation. In this presentation, we introduce a novel garment handling system based on garment features. The system is designed to improve the efficiency of grasping and unfolding T-shirts. The strategy utilizes a learning-based scheme and suggest candidate pairs of grasping points suitable for a dual-arm manipulator system. A pair of grasping points is then determined using our innovative scheme. The criteria for selecting a pair of grasping points is initially established through human demonstrations and is updated based on the results of robot operations to reflect the robot's performance in the garment handling. In addition, our approach is developed using real data rather than simulated environments. Experimental results show the effectiveness of the proposed strategy and demonstrate significant advances in robotic garment manipulation.