Towards Finding Optimal Solutions For Constrained Warehouse Layouts Using Answer Set Programming

Abstract

A minimum requirement of feasible order picking layouts is the accessibility of every storage location. Obeying only this requirement typically leads to a vast amount of different layouts that are theoretically possible. Being able to generate all of these layouts automatically opens the door for new layouts and is valuable training data for reinforcement learning, e.g., for operating strategies of automated guided vehicles. We propose an approach using answer set programming that is able to generate and select optimal order picking layouts with regards to a defined objective function for given warehouse structures in a short amount of time. This constitutes a significant step towards reliable artificial intelligence. In a first step all feasible layout solutions are generated and in a second step an objective function is applied to get an optimal layout with regards to a defined layout problem. In brownfield projects this can lead to non-traditional layouts that are manually hard to find. The implementation can be customized for different use cases in the field of order picking layout generation, while the core logic stays the same.