基于多Agent的多无人机协同决策算法仿真平台设计  

Simulation Platform of Cooperative Algorithm for UAVs Based on Multi-Agent System

作　　者：周昊[1] 覃征[1,2] 邢剑宽[2] 

机构地区：[1]清华大学软件学院,北京100084 [2]清华大学计算机科学与技术系,北京100084

出　　处：《系统仿真学报》2012年第3期587-593,共7页Journal of System Simulation

基　　金：国防"十一五"预研基金(402040202)

摘　　要：无人机协同作战的相关算法概括起来可以分为两类:一种是用于为无人机指明"往哪飞"的任务分配相关算法,另一类是用于确定"如何飞"的路径规划相关算法。首先针对这两大类协同算法归纳出相应的抽象数学模型,然后以此为依据确定无人机Agent的各决策模块的内部结构和彼此之间的交互接口,以及对外提供的扩展"钩子";然后在此基础上,利用多Agent技术搭建了无人机协同作战算法仿真平台UAVSim,以模拟从任务分配到路径规划的协同全过程。整个平台采用分层的体系结构,将算法、内核和显示完全分离。平台中参与模拟的无人机个体都用单独的Agent表示,每个Agent遵循统一的协作框架,同时可根据特定的协同算法动态配置其决策模块。Generally speaking,there are two kinds of algorithms for cooperative Unmanned Aerial Vehicles（UAVs） in battle scenarios： one is about task allocation,which indicates ＂where to fly＂,while the other is about path planning,which explains ＂how to fly＂.Firstly,the mathematical models for these two problems were established by analyzing their collaboration flows and picking up the commonalities and characteristics among these algorithms.Upon these models,secondly,the details of the decision-making module for each UAV was designed,which provided hooks for expanded features.Finally,the general simulation platform – UAVSim based on multi-agent technology was implemented,in which each UAV entity existed as an agent which can sense,made decisions according to the specified algorithm and react during the simulation,by following the uniform framework for collaboration.A layered architecture was adopted in this platform,which separated algorithms library and displayed from the core simulation.

关 键 词：无人机 仿真 协同 多AGENT 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]