The layout problem has been a focus point of research in factory planning for over six decades. Several newly emerging techniques for example genetic algorithms have been applied to the problem to generate better solutions closer to practical application. Nevertheless, solving the layout problem without considerable simplification of the base problem still presents a challenge. This publication shows how to model the layout problem in the framework of Markov decision processes (MDP) to apply reinforcement learning as a novel approach for generating layouts. Reinforcement learning (RF) has previously not been applied to the layout problem to the best knowledge of the author. Research in other fields of study shows the enormous potential of RF and the capability to reach superhuman performance in a variety of tasks. Although RF may not provide a better solution for finding the global optimum in the layout problem than genetic algorithms or dynamic programming, we hope to be able to include more constrains that matter for real world planning applications while keeping the calculation time feasibility short for practical application.