A Standardized Approach to Evaluate Assistive Systems for Manual Assembly Tasks in Industry

Abstract

Various assistive systems for supporting manual assembly tasks in industry have been proposed within the last years. These systems range from simple work instructions on screens, to animated mixed reality and in-situ projected hints directly onto the work-piece. Several design approaches exist for these systems. Commonly, assistive systems are tested and evaluated using abstract use cases with widely available building blocks (e.g. Duplo or Lego). These approaches work to test simple picking and placing tasks, but cannot reproduce the complexity of real assembly, which may include the tightening of screws, adding components from specific directions or attaching cables in a predefined way. Other evaluation approaches make use of realistic work processes from real industrial contexts. However, these evaluations are less controllable and hardly replicable. Our work presents a realistic and easyto-recreate approach to evaluate assistive systems for manual assembly tasks. It comprises a 3D-modelled object that considers several procedures and aspects of manual assembly and, as an abstract assembly object, helps to improve the usability of laboratory developed assistive systems. The model itself is developed to be 3D printed and uses standardized components to make it easily available for research and learning factories around the globe. With this approach, we want to support researchers in comparing and evaluating assistive systems for industrial assembly tasks by providing a realistic and easy-to-use test setup with high internal and external validity.