A Deep Reinforcement Learning-Driven Seagull Optimization Algorithm for Solving Multi-UAV Task Allocation Problem in Plateau Ecological Restoration

The rapid advancement of unmanned aerial vehicle (UAV) technology has enabled the coordinated operation of multi-UAV systems, offering significant applications in agriculture, logistics, environmental monitoring, and disaster relief. In agriculture, UAVs are widely utilized for tasks such as ecologi...

Full description

Saved in:
Bibliographic Details
Main Authors: Lijing Qin, Zhao Zhou, Huan Liu, Zhengang Yan, Yongqiang Dai
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Drones
Subjects:
Online Access:https://www.mdpi.com/2504-446X/9/6/436
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The rapid advancement of unmanned aerial vehicle (UAV) technology has enabled the coordinated operation of multi-UAV systems, offering significant applications in agriculture, logistics, environmental monitoring, and disaster relief. In agriculture, UAVs are widely utilized for tasks such as ecological restoration, crop monitoring, and fertilization, providing efficient and cost-effective solutions for improved productivity and sustainability. This study addresses the collaborative task allocation problem for multi-UAV systems, using ecological grassland restoration as a case study. A multi-objective, multi-constraint collaborative task allocation problem (MOMCCTAP) model was developed, incorporating constraints such as UAV collaboration, task completion priorities, and maximum range restrictions. The optimization objectives include minimizing the maximum task completion time for any UAV and minimizing the total time for all UAVs. To solve this model, a deep reinforcement learning-based seagull optimization algorithm (DRL-SOA) is proposed, which integrates deep reinforcement learning with the seagull optimization algorithm (SOA) for adaptive optimization. The algorithm improves both global and local search capabilities by optimizing key phases of seagull migration, attack, and post-attack refinement. Evaluation against five advanced swarm intelligence algorithms demonstrates that the DRL-SOA outperforms the alternatives in convergence speed and solution diversity, validating its efficacy for solving the MOMCCTAP.
ISSN:2504-446X