Advanced Optimisation in Freight Transport: Prospects and Challenges for Neural Combinatorial Optimisation for Integrated Transport Logistics Problems

Abstract

To ensure the attractiveness of intermodal freight transport (IFT), a large number of optimisation measures are needed. This involves numerous control decisions on different time scales, making integrated optimisation with traditional methods almost unfeasible. Recently, a new trend in optimisation science has emerged: neural combinatorial optimisation (NCO). NCO enables real-time decision making under uncertainties - a crucial factor for a seamless process synchronisation and consequently for the competitiveness of IFT. The objective in this study is twofold. First, we investigate optimisation problems in intermodal operations. Second, we thoroughly examine the current State of the Art in NCO, mapping the advanced optimisation techniques to the integrated problems in IFT. In particular, we notice that end-to-end techniques are widely used for operational problems such as scheduling and vehicle routing. We highlight the importance of Hierarchical Reinforcement Learning (HRL) and hybrid approaches for integrated decision making. Overall, despite the fact that NCO is currently in its infancy, this study advocates for its great potential in transport logistics.