Study Proposes Two-Stage UAV Framework for Urban Logistics and Flight Safety

Low-altitude UAV logistics networks in complex urban environments gain coordination through a collaborative two-stage optimization framework combining an improved Double Deep Q-Network with a risk-weighted Theta* algorithm to balance service loads and flight safety in Wuhan.

Urban delivery networks often struggle to coordinate transfer-node deployment with safe path planning. Dense populations and tall buildings complicate low-altitude airspace management. The new research addresses this coordination gap by merging macro-level node allocation with micro-level path generation.

Altitude-specific risk maps are constructed by incorporating population density, building height, and no-fly zones. These maps form the foundation for the upper-stage optimization process.

In the upper stage, a Double Deep Q-Network enhanced with NoisyNet and prioritized experience replay optimizes transfer-node selection and service allocation. The full model achieves a 22.3% boost in mean final reward and a 23.8% drop in mean episodes required to reach convergence relative to the standard DDQN.

The lower stage utilizes a risk-weighted Theta* algorithm to generate flight routes. This stage feeds path distances and risk costs back to the upper-level decisions.

Compared with conventional A*, risk-weighted Theta* reduces total service distance by 10.5%. It also cuts risk cost by 20.1% and total cost by 16.6%.

Metric Improvement
Mean Final Reward (vs. baseline DDQN) +22.3%
Episodes to Convergence (vs. baseline DDQN) -23.8%
Total Service Distance (vs. conventional A*) -10.5%
Risk Cost (vs. conventional A*) -20.1%
Total Cost (vs. conventional A*) -16.6%

Researchers conducted a case study in Wuhan to validate the framework. Results show that the developed architecture successfully harmonizes network configuration and risk-conscious path planning while simultaneously managing flight safety, service-load equity, and economic viability.

Wuhan, the capital of Hubei province with a population of 13,861,900, lies at the confluence of the Yangtze and Han rivers, a geography that historically earned it the name “Nine Provinces’ Thoroughfare.” Formed by the 1927 merger of Wuchang, Hankou, and Hanyang, the city has long functioned as a critical transportation hub where dense urban development and complex riverine topography create challenging low-altitude airspace conditions.

These results provide methodological support for UAV logistics network planning in complex urban environments.

Photo of author

Technology Desk

Technology Desk at Archyde.

Jared Padalecki Open to ‘Protection (Très) Rapprochée’ Sequel, Jokes About Role Reversal

Italy’s Supreme Court Upholds Life Sentence for Ex-Doctor Who Killed Wife, Mother-in-Law with Drug Cocktail