A multi-objective optimization framework for integrating drone-based urban air mobility with ground logistics networks: toward sustainable and resilient last-mile delivery

Rising urban population growth, congestion, and emissions pressures motivate urban air mobility (UAM) as a three-dimensional complement to ground logistics. This paper develops MO-UAMLN, a multi-objective mixed-integer linear program that jointly minimizes cost and emissions and maximizes disruption resilience for an integrated drone-ground vehicle network with vertiport infrastructure, solved via a novel Deep Reinforcement Learning-enhanced NSGA-III (DRL-NSGA-III) algorithm. In a Tehran case study (150 nodes, three distribution centers, eight vertiports), the integrated strategy cuts cost by 18.7% and emissions by 34.2% and raises expected disruption coverage to 88.0% versus single-mode baselines; DRL-NSGA-III significantly outperforms NSGA-II, NSGA-III, and MOEA/D (Holm-Bonferroni corrected p < 0.05, large effect sizes), and battery capacity, vertiport throughput, and demand density emerge as the dominant performance drivers. These results indicate that resilience arises from modal complementarity rather than redundant capacity, offering differentiated investment guidance for operators, planners, and policymakers. Findings rest on a single metropolitan case study and stylized assumptions, so generalization to other cities and real deployment conditions requires further validation.