跳到主要导航 跳到搜索 跳到主要内容

Reinforcement learning based dynamic allocation strategy for relief supplies by heterogeneous unmanned system

  • Northwestern Polytechnical University Xian
  • Beijing Institute of Tracking and Telecommunications Technology

科研成果: 期刊稿件 › 文章 › 同行评审

摘要

In this paper, the dynamic allocation problem of relief supplies by a heterogeneous unmanned system (HUS) is discussed. A reinforcement learning-based intelligent allocation strategy is proposed that considers dynamic changes in disaster areas and payload differences among heterogeneous unmanned platforms, which effectively addresses the allocation conundrum. Firstly, an evaluation function closely related to the elapsed time of the HUS is designed, enabling the quantitative assessment of resource allocation schemes, thereby providing a basis for optimizing allocation strategies. Secondly, in response to the varying velocities and load capacities among HUS, an intelligent task allocation strategy is specifically devised to accommodate rescue operations. In addition, variations are introduced in both the number of affected areas and the demand for supplies. The application of intelligent algorithms enables rescue operations to consistently maintain efficient and flexible execution capabilities across various scenarios through dynamic adjustment strategies. Finally, compared to the strategy optimized by traditional distance-based optimization methods and greedy algorithms, the strategy generated by the proposed algorithm reduces task completion time and demonstrates better performance in rescue assignment.

源语言英语
期刊论文编号399
期刊Journal of King Saud University - Computer and Information Sciences
卷38
期6
DOI
出版状态已出版 - 8月 2026

学术指纹

探究 'Reinforcement learning based dynamic allocation strategy for relief supplies by heterogeneous unmanned system' 的科研主题。它们共同构成独一无二的学术指纹。

引用此