基于深度强化学习的多无人机任务动态分配  

Dynamic Task Allocation Method for UAVs Based on Deep Reinforcement Learning

作　　者：唐峯竹 唐欣 李春海 李晓欢[1] TANG Fengzhu;TANG Xin;LI Chunhai;LI Xiaohuan(School of Information and Communication,Guilin University of Electronic Technology,Guilin Guangxi 541004,China;Institute of Information Technology of GUET,Guilin Guangxi 541004,China)

机构地区：[1]桂林电子科技大学信息与通信学院,广西桂林541004 [2]桂林电子科技大学信息科技学院,广西桂林541004

出　　处：《广西师范大学学报（自然科学版）》2021年第6期63-71,共9页Journal of Guangxi Normal University:Natural Science Edition

基　　金：国家自然科学基金(61762030);广西自然科学基金(2019GXNSFFA245007,2018GXNSFDA281013);广西科技计划项目(AA18242021,AB19110050,AA19110044,ZY19183005,AB20238033);桂林市科技计划项目(20190214-3);广西高校中青年教师基础能力提升项目(2021KY1654)。

摘　　要：针对无人机任务随机下发场景中由于任务完成时间约束带来的任务完成度低的问题,本文提出一种基于深度强化学习的分布式多无人机任务动态分配方法。该方法通过无人机之间的实时交互,首先,对正在执行的任务和新到任务的时间约束、任务量、优先级等特征进行实时量化,同时在任务执行过程中动态生成新的任务优先级特征;然后,将无人机交互后形成的全局任务特征视为无人机共享的任务情况,并不断进行更新形成动态决策依据;最后,在时间约束下,根据实时任务完成情况通过深度强化学习方法进行无人机的行为决策,达到新任务与正在执行任务的动态分配以提高任务完成度。仿真结果表明,该方法能提高时间约束下的系统整体任务完成度。Aiming at the problem of low task completion caused by task completion time constraints in the scenario where tasks are randomly assigned,a distributed and dynamic multi-UAV task allocation method based on deep reinforcement learning is proposed.The method uses the interaction between UAVs to quantify the time constraints,task size,task priority and other characteristics of the tasks being performed and new tasks in real time.At the same time,new task priority features are generated dynamically during task execution.Then,the task features after UAV interaction are regarded as the global task shared by UAV,and constantly updated to form a dynamic decision-making basis.Finally,according to the real-time task completion and the time constraints,the behavioral decision-making of the UAV is made based on the deep reinforcement learning,so as to improve the task completion by achieving the dynamic allocation of new tasks and ongoing tasks.This behavioral decision is to realize the dynamic assignment of tasks to improve the task completion.Simulation results show that this method can improve the overall task completion of the system under time constraints.

关 键 词：无人机 任务分配 深度强化学习 动态性 任务完成度 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]