Enhancing UAV Safety in Smart City Environments: A Dynamic Constrained Multi-Objective Evolutionary Algorithm

The integration of Unmanned Aerial Vehicles (UAVs) as mobile edge computing nodes has become increasingly crucial in smart city environments, where they provide dynamic computational support and services in areas with sudden high computational demands. Effective and safe UAV path planning is essential to ensure these aerial platforms can navigate complex urban landscapes while minimizing risks to ground infrastructure and populations. Current UAV path planning problems are commonly modeled as constrained multi-objective optimization problems, which typically focus on static environmental conditions and fail to adequately address the dynamic changes in safety levels during flight in real-world urban scenarios. These static models are insufficient for real-world urban scenarios where environmental conditions constantly fluctuate due to changing densities, weather conditions, and other dynamic factors. To address this, this paper proposes a dynamic constrained multiobjective evolutionary algorithm based on a population generation mechanism. First, a dynamic constrained multi-objective optimization model is established, incorporating environmental constraints, performance constraints, and the dynamic variations in environmental conditions during UAV flight. Second, a population generation mechanism is designed to rapidly generate feasible initial solutions adapted to new environmental conditions after dynamic changes, significantly improving the proposed algorithm’s adaptability to real-time variations and planning efficiency. Simulation experiments comparing the proposed algorithm with three baseline algorithms demonstrate significant improvements in planning success rate and solution set quality.