Machine Learning in Production – Potentials, Challenges and Exemplary Applications

TLDR

Machine learning has driven advances in autonomous driving, NLP, robotics, and Industry 4.0, yet the manufacturing sector still lacks a systematic, economically viable guideline and structured overview of industry‑specific best practices for deploying ML. This paper compiles existing ML application scenarios across manufacturing processes and industry sectors to provide a comprehensive reference. The authors map scenarios to DIN 8580 process groups, VDI 2860 handling operations, and cross‑process approaches, and illustrate sector‑specific examples from electronics, electric motors, transmission components, and medical devices.

Abstract

Recent trends like autonomous driving, natural language processing, service robotics or Industry 4.0 are mainly based on the tremendous progress made in the field of machine learning (ML). The increased data availability coupled with affordable computing power and easy-to-use software tools have laid the foundation for using such algorithms in a wide range of industrial applications, e.g. for predictive maintenance, predictive quality or machine vision. However, a systematic guideline for identifying and implementing economically viable ML use cases in manufacturing industry is still missing. In particular, there is still a lack of a structured overview of concrete, industry-specific best practices that can be easily transferred to one’ s own production. Hence, this paper aims to summarize various existing application scenarios of ML from a process and an industry sector perspective. The process point of view mainly covers the main manufacturing process groups of DIN 8580, handling operations according to VDI 2860 as well as selected cross-process approaches. From an industry sector perspective, application scenarios from various subsectors such as the production of electronics, electric motors, transmission components and medical devices are outlined.