A Generic Modelling Framework for Last-Mile Delivery Systems

TL;DR

The paper tackles the complexity of urban last-mile delivery by presenting a generic, modular modelling framework that can be calibrated to different cities and integrated with existing simulation tools. It combines HUMAT, a socio-cognitive ABM, with MASS-GT, a freight-focused ABM, in a two-phase Setup and Execution architecture that ensures data compatibility and transferability. Two real-world case studies—crowdshipping in The Hague and parcel lockers in Thessaloniki—demonstrate the framework's adaptability and reveal how consumer behavior influences delivery efficiency, congestion, and emissions. The work advances computational social science for urban logistics by offering a scalable tool for planners and policymakers, with future prospects including drones, AI-driven routing, and expanded socio-economic analyses under the URBANE Horizon project.

Abstract

Large-scale social digital twinning projects are complex with multiple objectives. For example, a social digital twinning platform for innovative last-mile delivery solutions may aim to assess consumer delivery method choices within their social environment. However, no single tool can achieve all objectives. Different simulators exist for consumer behavior and freight transport. Therefore, we propose a high-level architecture and present a blueprint for a generic modelling framework. This includes defining modules, input/output data, and interconnections, while addressing data suitability and compatibility risks. We demonstrate the framework's effectiveness with two real-world case studies.

Figures (5)

• Figure 1: HUMAT - Setup phase
• Figure 2: MASS-GT modules
• Figure 3: Integrated generic modelling framework - all phases
• Figure 4: Modelling framework calibrated for crowdshipping.
• Figure 5: Modelling framework calibrated for parcel locker service.