The current shortage of road and parking capacity to accommodate freight traffic poses a significant challenge in cities. This study develops and analyzes alternative traffic management strategies for last-mile delivery operations. Three alternative implementations of parking guidance involving allocating commercial vehicles to dedicated loading/unloading bays are investigated alongside a vehicle-specific geofence strategy. Methodologically, an agent-based model framework is employed to reproduce the interactions among (parking and cruising) carriers, the surrounding traffic, and a traffic controller. An efficient metaheuristic is integrated with simulation to address the corresponding optimization. The effectiveness of the strategies in reducing traffic congestion and other externalities varies depending on the level and configuration of freight demand. Among the parking guidance strategies, those weighing more on carriers’ convenience mitigate potential risks of equity and acceptability issues but at the cost of an efficiency loss. Geofencing is less problematic due to the minor operational modifications, offering comparable traffic performance improvement for low and medium demand levels.
Parking Guidance and Geofencing for Last-Mile Delivery Operations
IEEE Transactions on Intelligent Transportation Systems ; 25 , 8 ; 9091-9102
01.08.2024
7677656 byte
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
A reinforcement learning framework for improving parking decisions in last-mile delivery
Taylor & Francis Verlag | 2024
|