多智能主体协作规划理论及其在RoboCup中的应用  

Multi-Agent Collaborative Planning and Its Application in Robocup

作　　者：杜陶钧[1] 黄鸿[1] 

机构地区：[1]北京理工大学信息工程学院,北京100081

出　　处：《计算机仿真》2004年第7期125-128,共4页Computer Simulation

摘　　要：机器人足球赛是近年人工智能迅速发展起来的一个重要的研究领域。基于此平台,研究了将规划、竞争、约束、协调纳入一个协作规划框架之中的多智能体协作规划理论,并将其分为5个阶段①产生协作目标;②协作规划;③寻求协作伙伴;④选择协作方案;⑤实现所确定的目标。最后探讨了应用到机器人足球赛的一些技术问题。在比赛中,决策速度也较好。The Robocup soccer tournament, which has fast developed in recent years , is a very important research area in AI. In this paper, cooperation process is studied by bringing planning competition, constraints, and coordination into a cooperation framework. In the framework, cooperation is divided into six stages: ①producing cooperative goal ,②collaborative planning, ③looking for cooperation partner,④selecting solution,⑤achieving goal. Finally the application of the framework to Robocup is also studied in the paper. In Robocup race, decision making of our team is faster.

关 键 词：机器人足球赛 人工智能 多智能体协作规划 协作伙伴 形式化框架 多主体协作系统 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]