Anticipatory freight selection in intermodal long-haul round-trips

We consider the planning problem faced by Logistic Service Providers (LSPs) transporting freights periodically, using long-haul round-trips. In each round-trip, freights are delivered and picked up at different locations within one region. Freights have time-windows and become known gradually over time. Using probabilistic knowledge about future freights, the LSP's objective is to minimize costs over a multi-period horizon. We propose a look-ahead planning method using Approximate Dynamic Programming. Experiments show that our approach reduces costs up to 25.5% compared to a single-period optimization approach. We provide managerial insights for several intermodal long-haul round-trips settings and provide directions for further research.