Two-Stage Framework Optimizes Urban UAV Logistics and Risk-Aware Routing
A study addresses insufficient coordination between transfer-node deployment and risk-aware route planning in complex urban environments using a collaborative optimization framework.
The Tech TL;DR:
- A new two-stage framework coordinates drone transfer nodes and flight paths using an improved Double Deep Q-Network and a risk-weighted Theta* algorithm.
- Altitude-specific risk maps factor in building heights, population density, and restricted airspace.
- Tests show the framework increases final reward by 22.3% and reduces total logistics costs by 16.6% compared to baseline methods.
Urban drone logistics often suffer from poor coordination between where delivery nodes are placed and how drones route around obstacles. To fix this bottleneck, a newly proposed two-stage collaborative optimization framework integrates an improved Double Deep Q-Network with a risk-weighted Theta* algorithm.
Constructing altitude-specific risk maps forms the foundation of the approach, incorporating population density, building height, and no-fly zones. In the upper stage of the model, the DDQN enhanced with NoisyNet and prioritized experience replay optimizes transfer-node selection and service allocation.
Within the lower tier, path distances along with risk expenses are computed by the risk-weighted Theta* algorithm and returned to inform the decisions made at the upper level. A case study in Wuhan was conducted to validate the framework.
When evaluated against standard DDQN, the full system achieves a 22.3% boost in average final reward alongside a 23.8% decrease in average episodes needed for convergence.
Relative to traditional A*, the risk-weighted Theta* method cuts total service distance, risk expenditure, and overall expenses by 10.5%, 20.1%, and 16.6%, respectively. Ultimately, these results show that the suggested model successfully aligns network structure with risk-conscious pathfinding while maintaining a trade-off between economic viability, workload distribution, and safe flight operations, offering valuable methodological backing for designing UAV delivery systems across dense cityscapes.
Wuhan, the capital of Hubei with a population of 13,861,900, lies at the confluence of the Yangtze and Han rivers, a geography that earned it the nickname “Nine Provinces’ Thoroughfare.” Its urban area is divided into the three historic towns of Wuchang, Hankou, and Hanyang by these waterways, creating a dense, complex cityscape that makes it a representative testbed for low-altitude logistics networks.
Always consult with certified IT and cybersecurity professionals before altering enterprise networks or handling sensitive data.