Construction of Bin-Picking System for Logistic Application: A Hybrid Robotic Gripper and Vision-Based Grasp Planning

Abstract

An autonomous bin-picking system for grasping various cluttered packages can significantly benefit logistics by reducing manual labor and streamlining processing. The system's key challenges involve the gripper for confined spaces and grasp planning for unseen objects with varying materials, shapes, and sizes. To address these issues effectively, we propose a bin-picking system that includes a novel gripper and a corresponding vision-based grasp planning strategy. Firstly, a multi-mode hybrid gripper combining suction and pinch is developed to enhance versatility, as pinch alone fails for oversize objects and suction struggles with uneven surfaces. By integrating the suction cup into a slender finger and employing a flipping module and underactuated linkages, the compactness and dexterity are enhanced, ensuring the handling of packages near the bin walls or corners. Secondly, a model-free heuristic grasp planning framework based on the unseen object instance segmentation (UOIS) network is designed for grasping packages in a cluttered bin, which can be applied to hybrid grippers. Thirdly, we compared the prototype's hardware characteristics with Hand-E and conducted grasping experiments to demonstrate the functionalities of the proposed hybrid gripper. Finally, the autonomous package bin-picking system was evaluated in a simulator, achieving a 71.4% success rate, compared to mono-functional grippers such as suction (53.9%) and Hand-E (39.3%). Real-world experiments further validated its practicality, highlighting its potential in logistics scenarios.