A Framework for Automated Multiobjective Factory Layout Planning using Reinforcement Learning

Abstract

Layout planning is a central element of the factory planning process. Given its complexity, layout planning is often time consuming and involves creative processes. One possible way to deal with this complexity is the training of a machine learning algorithm, which enables to generate and optimize factory layouts. Consequently, this paper outlines a reinforcement learning based concept for automated layout planning. In particular, boundary conditions and objective functions are derived from the existing planning parameters of factory layouts. The presented approach will allow a multiobjective optimization of the layout and uses material flow and energy consumption as optimization criteria.