3D Shape Tracking of Deformable Linear Objects via Tactile-Visual Fusion for Collaborative Assembly

KejiaChenTUM

Attend In Person: Yes

Keywords: Deformable Linear Objects, Shape Tracking, Assembly

TL;DR: We present a tactile-visual fusion method for continuous 3D shape tracking of deformable linear objects (DLOs) during robotic manipulation.

Abstract: Assembly of deformable linear objects (DLOs) such as cables with robots requires accurate tracking of its 3D shape. In this work, we propose a tactile-visual fusion framework for continuous 3D shape tracking during robotic manipulation. Our approach integrates global 3D visual estimation with local tactile sensing from ViTac-equipped grippers to recover occluded shape information and refine global tracking. The fused estimation supports accurate manipulation for collaborative multi-robot tasks. We validate the method in an inter-robot handover scenario, where one robot transfers a grasped DLO and hands it precisely to another. Results demonstrate improved shape tracking accuracy and robust performance under occlusions.

Submission Number: 25