Optimizing UAV last-mile delivery with constraint-aware evolutionary computation

초록

Unmanned aerial vehicles (UAVs) have demonstrated great potential for last-mile package delivery due to their accessibility, efficiency, cost-effectiveness, environmental advantages, and ability to provide contactless services. Nevertheless, significant challenges remain, including the avoidance of no-fly zones, the handling of fragile cargo, and strict motion constraints such as limits on speed, acceleration, and safe takeoff/landing requirements. To address these challenges, this paper proposes a last-mile delivery model (LMDM) that incorporates no-fly zone avoidance, fragile cargo protection, motion constraints, and energy consumption minimization into UAV trajectory planning. Since the LMDM is inherently non-convex and difficult to solve using conventional optimization methods, we employ evolutionary computation (EC) techniques, along with a constraint-handling mechanism designed to ensure feasibility while achieving energy-efficient UAV trajectories. Numerical evaluations demonstrate that the proposed constraint-handling method can efficiently produce feasible solutions that satisfy various constraints of last-mile delivery and reduce total energy consumption, whereas the conventional penalty function approach fails to find feasible ones. © 2026 The Authors.

키워드