Four Independent Knowledge Domains to Enable an Agile, Distributed Development of User-Centred Engineering Configurators

Abstract

Robot-centric automation solutions (RAS) promise greater efficiency and consistent quality in production, relieving workers of physically demanding and dangerous tasks, especially in the times of COVID-19. Nevertheless, due to their relatively high complexity and implementation costs, RAS are only used to a limited extent by small and medium-sized manufacturing companies. As a rule, the high costs of RAS arise from custom engineering efforts, which take up to 70 percent of the acquisition costs. For this reason, it is necessary to optimise the engineering of RAS. However, software tools such as configurators have been used primarily for the individualisation of products, such as automobiles or clothing, based on variants predefined by the manufacturer, and less for the engineering of automation solutions. The development of knowledge-based systems, in particular knowledge-based engineering configurators (EC), is usually performed by few proficient experts with high development effort. One of the primary challenges in the knowledge acquisition is that several experts possess partial aspects of knowledge in an inhomogeneous, implicit form. Furthermore, there is a lack of efficient development methods for EC. By reusing knowledge elements from previous development projects, a sustainable increase in efficiency is possible. In order to enable an efficient development process of EC, we introduce a structuring model consisting of four knowledge domains (KD): knowledge about specific business cases (KD1), Best Practices as case-specific solution knowledge (KD2), logical expert knowledge (KD3) as well as semantically consistent data models for interoperability of different IT systems (KD4). As the four KD are independent, their development can be agilely divided among several teams or companies. Finally, the agile development approach is validated individually for each KD as well as comprehensively within the scope of the ROBOTOP platform for planning RAS.