Towards a user-centered development process of machine learning applications for manufacturing domain experts

The paradigm shift in manufacturing, the transition from mass production to mass customization [1], increases the engineering workload in planning and preparing the manufacturing process. The current software tooling for the engineering activities (computer-aided design, computer-aided manufacturing, etc.) are well suited for mass production, but not for mass customization. Artificial intelligence in general and novel developments in machine learning specifically have great potential in attenuating the workload increase of manufacturing engineers caused by the increasing product variance. One important barrier to consider in adopting AI-based software tooling is their acceptance by their targeted users. Manufacturing engineers usually lack in-depth AI know-how and cannot efficiently use AI or ML methods to reduce their workload, although their type of work would benefit greatly from the use of such methods. This paper explores the use of user-centered design concepts in the context of enabling manufacturing engineers to efficiently use AI and/or ML methods.