Partial Fabric Flattening for Seam Line Region Using Dual-Arm Manipulation

Robotic fabric flattening is essential for precision sewing. Existing methods primarily focus on an entire fabric flattening operation by the robot. This operation is inefficient for sewing and is also limited by the robot's workspace. In this paper, we propose a partial fabric flattening method using a dual-arm manipulator, specifically targeting regions around the seam line. Our method first employs a deep learning-based segmentation model to identify seam lines and wrinkles, and then selects the biggest wrinkle within a seam-aligned bounding box. A geometric pipeline determines the manipulation points, and a dual-arm manipulator performs symmetric pulling between the manipulation points using the impedance control method. Experimental results show that the proposed method effectively removes wrinkles around the seam line.