A Practical Solution to Deformable Linear Object Manipulation: A Case Study on Cable Harness Connection

The manipulation of deformable linear objects (DLOs) is one of the great challenges remaining in robot-centered flexible automation. Typical industrial applications of DLO manipulation include cable harness routing, bundling and assembly. DLOs usually have high degrees of freedom which makes it difficult in modelling, state estimation and operation, This paper focuses on one sub-category of DLO, the so-called semi-deformable linear object (SDLO), which consists of at least one rigid part. Taking cable harness-socket connection as a study case, a systematic solution is proposed to achieve successful manipulation of SDLOs. It consists of machine vision based 3D pose estimation and dual-arm robot based operation. The pose estimation is to figure out the position of the rigid part of a SDLO by learning-based object detection and to obtain its accurate pose by shape-based 3D matching. In the operation, strategies for dual-arm robot are developed to handle various initial poses of the SDLO and to realize connection task in all cases. The verification results show the proposed solution is feasible for manipulating SDLOs in a practical industrial scenario.