Optimization of Semi-Open Multi-Depot Vehicle Routing and Delivery Scheduling

Abstract

With the ongoing advancement of the "Internet Plus" initiative and the sharing economy, crowd-sourced delivery—a critical component of urban instant logistics—faces challenges such as high order heterogeneity and complex resource allocation. This study constructs a semi-open multi-depot scheduling model with dual objectives of minimizing delivery costs and maximizing customer satisfaction. The model systematically incorporates an order classification mechanism and a rider competency hierarchy, while integrating multiple real-world constraints including time windows, task load balancing, and route feasibility. To efficiently solve this NP-hard problem, we propose an improved genetic algorithm (IGA) featuring a dual-layer encoding structure. By introducing a coordinated optimization strategy for task assignment and route sequencing, the algorithm enhances solution stability and quality in complex scenarios. Numerical experiments based on real-world platform data demonstrate the superiority of the model and algorithm in scheduling efficiency, route optimization, and satisfaction improvement. This research provides both theoretical foundations and empirical evidence for intelligent scheduling in instant delivery platforms.