# Multi-objective optimization model for dynamic airspace allocation of urban drones in medical emergency delivery

## TL;DR

This paper introduces a new system using drones for emergency medical deliveries in cities, aiming to reduce response times and costs.

## Contribution

A novel multi-objective optimization model with real-time adaptability for urban drone emergency delivery is proposed.

## Key findings

- Response time decreased by 61% and delivery costs by 85% compared to ground-based services.
- The system maintained 85-96% performance in adverse conditions with adaptation cycles under 2 seconds.
- UAV fleet size was identified as the primary factor influencing system performance.

## Abstract

Urban emergency delivery systems face serious challenges in dealing with traffic congestion, structural limitations, and accessibility challenges that confront traditional ground-based emergency medical services. This study proposes a novel multi-objective optimization approach to the dynamic airspace allocation for unmanned aerial vehicles (UAVs) in urban emergency environments.

The approach simultaneously optimizes four competing objectives including the decrease in delivery time, the costs, the increase in airspace usage efficiency, and the reduction in risks incurred by hazards to safety. An Improved NSGA-III (Non-dominated Sorting Genetic Algorithm III) algorithm with real-time adaptability characteristics and conflict resolution algorithms for emergency conditions was developed. The effectiveness was evaluated through in-depth Monte Carlo simulations that applied real urban emergency scenarios from the metropolitan area of Hangzhou, taking measurements from a metropolitan area of 150 km2 with 35 tertiary hospitals, among different UAV fleet configurations.

Simulation results suggest considerable performance improvements, with a decrease in response time by 61% (from 18.5 to 7.2 min), as well as a decrease in delivery costs by 85% (from $280 to $42.3) when compared to calibrated ground-based emergency services baseline. The dynamic reallocation mechanism provided 85 to 96% reference performance in adverse conditions such as mass incidents, adverse weather, and equipment failures, with adaptation cycles achieved in average time of 1.8 ± 0.3 s. Sensitivity analyses revealed that the size of the UAV fleet emerged as the primary performance-influencing factor.

This simulation-based framework provides a foundation for future emergency response system improvements, with pilot trials and hardware-in-the-loop testing representing essential next steps for practical deployment. The findings have considerable policy implications for urban air mobility planning and resource allocation.