Green and Intelligent Planning of Drone Launch in Truck-Drone Collaborative Delivery

The Internet of Things (IoT) paradigm has enabled innovative applications across various domains, significantly enhancing efficiency in the transportation sector through intelligence-driven and sustainable solutions. In the field of parcel delivery, the integration of trucks and drones has attracted considerable attention from both academia and industry as a means to optimize logistics networks and reduce last-mile delivery costs. Traditionally, research on truck-drone collaborative delivery (TDCD) has focused on routing and scheduling problems within hypothetical scenarios. This study, however, seeks to address the problem using a more realistic approach by introducing a newly generated customer order dataset, which includes data from 191 customer locations over a span of 7 days. The goal is to evaluate the efficiency of drone deliveries assisted by trucks. Utilizing this dataset, we applied the Self-Organizing Feature Map (SOFM) algorithm, a type of artificial neural network, to the TDCD problem. This novel approach identifies the optimal location for truck-based drone launches to minimize overall travel distance. Thanks to its adaptive nature, the SOFM algorithm dynamically selects the launch location based on daily customer orders rather than relying on a static, predetermined site. This method has resulted in a 4.4% reduction in the total distance traveled by drones and a $\mathbf{1. 1 \%}$ reduction in the distance covered by trucks over the seven-day period. These efficiencies translate into savings of $30.28 \mathrm{gCO2}$ in carbon emissions and 80.16 Wh of energy consumption, equivalent to 288.58 Kjoules.