A novel drone delivery problem with stochastic arrival time under uncertain wind conditions for last-mile delivery using an iterated adaptive tabu search algorithm

With the rapid growth of e-commerce, last-mile delivery becomes a major challenge in logistics. Drones offer a flexible, cost-effective and environmental friendly solution. However, their energy consumption and arrival time are sensitive to environmental factors, e.g. uncertain wind conditions, making it challenging to offer punctual delivery services. Few studies have considered the stochasticity of wind effects on drone arrival time, which limits their applicability in real-world scenarios. This study establishes an extended chance-constrained programming (ECCP) model for a novel drone delivery problem with stochastic arrival time under uncertain wind conditions. A penalty cost is posed to consider customer satisfaction. A nonlinear energy function considering the wind effect is incorporated to better simulate real-world situations. To solve the ECCP model, a new iterated adaptive tabu search (IATS) algorithm with two-dimensional coding scheme is proposed, which incorporates tabu search within an iterative framework with multiple operators. A self-adaptive mechanism is applied to enhance search efficiency. Experimental results across 21 instances show that the proposed algorithm outperforms five baseline algorithms in solving the proposed model.