A Priority-Based Hybrid Algorithm for UAV Path Planning

In recent years, Unmanned Aerial Vehicles (UAVs) have been widely used for reconnaissance missions. Our main goal is to focus on high-value targets while reducing the flight time under UAV power limitations. We propose a framework named DQN-ALNS, which combines Deep Reinforcement Learning (DRL) with the Adaptive Large Neighborhood Search (ALNS) metaheuristic algorithm to improve the optimization process by exploring the current solution state. Specifically, the DRL agent is responsible for selecting the destroy-repair operator, enabling the construction of new solutions and iterative optimization of the UAV reconnaissance path. Experimental results show that DQN-ALNS produces superior solutions and converges faster than other benchmark algorithms. The algorithm leverages the exploratory potential of the current solution and demonstrates strong stability.