Machine Learning-Enhanced Cloud-Assisted Last-Mile Delivery Optimization with Priority Management Based on Queueing Theory

The rapid growth of e-commerce has escalated the complexity of last-mile delivery, challenging traditional static optimization methods. We propose the Machine Learning-Enhanced Cloud-Assisted Last-Mile Optimization (ML-CALMO) framework, which integrates deep learning, reinforcement learning, and queueing theory to address dynamic demand and priority orders. Utilizing a cloud-based infrastructure, ML-CALMO employs a dynamic vehicle-customer preference matrix for real-time routing optimization and predictive demand modeling. Experimental validation demonstrates a 28.3% reduction in average delivery time, 22.1% improvement in service efficiency, and 89.7% service success rate. Validated by continuous-time Markov chain analysis, ML-CALMO achieves high prediction accuracy with MAPE of 7.1% and deviates less than 5% from analytical results. This framework offers a scalable, industry-ready solution for modern logistics, enhancing efficiency and customer satisfaction.