Generation of design features by means of cluster analysis to develop a bioinspired picking gripper for intra-logistics

Abstract

Abstract: In the automation of intralogistical systems,it is discernible that order picking has a great potential for the future. Fundamental part of the automation process is the use of industrial robots that have to be equipped with a suitable end-effector, the gripper. The robot is capable of faster, more precise and more durable than the human pickers and thus contributes significantly to the efficiency increase. A major challenge for this development step for the substitution of manual picking is the design and provision of a suitable gripper system. With the experience of a previous research project, at the Chair of Machine Elements and Technical Logistics of the Helmut Schmidt University, the method of cluster analysis was first used to study the development of grasping objects, for a biomimetic universal gripper for picking drugstore products. This paper describes a contribution to the development of this gripper on the example of commercial drugstore products that are currently picked manually. These are clustered in terms of object features relevant for grasping and derived the resulting insights in the form of design features. After an analysis and definition of the relevant characteristics of grasping objects cross a database object is created. Using suitable methods based on the data obtained is processed and reduced. Following the gripping objects and their characteristic values of a hierarchical cluster analysis is carried out. Here, the boundaries of the clusters formed are determined by the associated gripping objects and analyzed. Finally, certain specific gripper features are reviewed and evaluated for applicability in the clusters. These considerations make it possible that special requirements on the gripper, which resulting directly from the properties of objects, these can reliably be constructive recognized and considered.