Transformation of robotic workcells to digital twins

Abstract

There is a growing demand for adaptive, off-line programmed robotic cells that can eliminate the laborious and time-consuming on-line programming work. Motivated by industrial requests, the paper proposes a design approach that transforms the already existing physical workcells to parametric digital twins. By improving twin closeness through multi-level calibration methods, an accurate digital twin of the as-built cell can be established, in which the sufficient accuracy of the off-line planned robotic operations is ensured. The approach is presented through the design of a belt grinding and polishing robotic cell for a cast aluminium workpiece in a real industrial scenario.

Introduction

Driven by industrial digitalization, the concept of Digital Twin (DT) has gained significant attention in the past few years. With the technological development in the fields of manufacturing, communication and information, the application of DTs became enabled [1]. As being a replica of the physical world, in addition to simulation, DTs allow optimization, prediction, observation and control on their real counterpart [2,3].

Although virtual validation is part of the DT concept, in case of robotic production systems, there is not much emphasis on physical verification and feedback of system commissioning [4], [5], [6]. Due to the inevitable deviations between the as-designed and as-built cells [7], DT based physical implementation may fail, requiring revision of the digital and/or the physical counterpart.

The present paper focuses on the model preparation aspect of DTs considering already existing physical workcells. The proposed kinematic linkage modelling technique aims to overcome the geometry and tolerance related implementation issues, by iteratively refining the twin closeness between the digital and the physical twin to ensure commissioning feasibility. This way, the prepared DT allows the application of off-line techniques (executed on the digital system) like off-line programming (OLP) or layout optimization in case of a robotic workcell [8], while minimizing on-line intervention (i.e., work in the real workcell). Moreover, as a programming, simulation and diagnostic tool, the DT supports the workcell throughout the system lifecycle.