Speaker: Max Cubillos Alvarez, Politecnico di Milano

Title: Truck scheduling for yard management using Estimated Time of Arrival

Abstract: Abstract: In the transportation sector, technology advances and new information systems have made available accurate real-time predictions of the Estimated Time of Arrival (ETA) of vehicles. In this study, we focus on the synchronization of the use of ETA for trucks into the yard management problem, which entails the assignment of trucks to docks in large warehouses with uncertain truck arrival times. We aim to quantify the added value of the use of ETA to generate schedules that minimize the waiting time of trucks.  For this, we propose a Markov Decision Process (MDP) formulation of the problem. To overcome the explosion in the dimension of the state and action spaces, we propose a heuristic solution based on a roll-out algorithm. We propose an Iterated Local Search (ILS) heuristic with a Variable Neighborhood Descent (VND) search to efficiently select the underlying scheduling actions in the model. We perform computational experiments to demonstrate the efficiency of our algorithm and compare the resulting policies with other benchmark scheduling rules used in parallel machine scheduling problems. Also, we assess the effect of different ETA modeling assumptions regarding distributional changes over time.

Coordinators: Marcel Turkensteen and Ata Jalili Marand