As cities worldwide begin embracing low-altitude logistics to support rapid, flexible deliveries by drones, urban planners face an increasingly difficult challenge: how to design an aerial delivery network that balances cost efficiency, safety, and noise impact.
A research team from Beihang University has developed a new framework that tackles this challenge head-on. Their study presents a multi-layered, hub-and-spoke logistics network design optimized using a dual-population co-evolutionary algorithm. This method not only improves route planning and facility placement but also explicitly accounts for noise constraints — a key concern for residents living near hospitals, schools, and housing complexes.
The team published their work in Acta Aeronautica et Astronautica Sinica on March 12, 2025.
"Noise is often the neglected factor in low-altitude logistics," said Dr. Yumeng Li, the corresponding author and associate professor at Beihang University. "But as drone delivery scales up, the public's tolerance for frequent overflights will be tested. Our approach takes noise considerations into the core of the logistics network's design."
A key innovation is the adoption of a multi-layered hub-and-spoke network structure, where nodes (takeoff/landing points and delivery centers) are mapped onto different altitude levels, and links represent horizontal drone flight paths. The model incorporates population-density-based risk zoning and classifies noise-sensitive facilities such as residential areas as constraints.
To quantify and constrain the noise impact, the researchers developed a three-dimensional noise exposure and equivalent sound level model. Noise from individual drone flights is measured using exposure level calculations, while the cumulative effect of multiple flights over time is captured using equivalent continuous sound levels. These metrics are then used to define noise tolerance thresholds for sensitive locations in the network.
The resulting multi-objective mixed-integer programming model includes constraints for network structure, traffic assignment, drone range, and noise limits. Solving this model is non-trivial due to its high dimensionality and complex constraints.
To address the computational challenge, the team proposed a Dual-Population Cooperative Co-evolutionary Algorithm for Constrained Multiobjective Optimization with Adaptive Operator Selection (CCMO-AOS). This algorithm maintains two populations: a main population that evolves under noise constraints, and an auxiliary population that explores the solution space without noise constraints to improve diversity and avoid local optima. An adaptive operator selection mechanism dynamically chooses suitable crossover and mutation strategies based on performance, while solution exchange between populations facilitates knowledge transfer and search space exploration.
Experimental results demonstrate that the proposed method effectively reduces noise exposure in sensitive areas while maintaining competitive performance in cost and safety. The network topology generated under noise constraints significantly differs from that without such constraints, illustrating the necessity of considering acoustic impact in network design.
This work provides not only a feasible technical pathway for integrating noise control into urban drone logistics planning, but also offers theoretical insights into solving constrained multi-objective problems in irregular search spaces. In future work, the researchers plan to incorporate dynamic and uncertain elements into the problem formulation, making the model more applicable to real-world scenarios.
Other contributors include Chunxiao Zhang and Tong Guo from Beihang University.
Original Source
HANG C X，GUO T，LI Y M. Dual-population coevolutionary optimization for multi-layer urban air logistics network [J]. Acta Aeronautica et Astronautica Sinica， 2025, 46(11): 531477 (in Chinese). doi：10. 7527/S1000-6893. 2024. 31477.
