摘要
机器人足球比赛是一个有意义而且复杂的新兴的人工智能研究领域,它是一个典型的多智能体系统。RoboCup是一个通用的实验平台,在该平台上可以评价各种理论和算法。本文采基于Q算法的激励学习研究了机器人的截球技术,将其应用在RoboCup仿真平台上,智能体经过多次的学习之后,实验表明截球技术取得了理想的效果。
Robot Soccer is an interesting and complex new area of artificial intelligence,and it is a typical multi-Agent system.RoboCup is a common experimental platform by which theories and algorithms can be evaluated.This paper descries the use of reinforcement learning base on Q algorithm for robot interception technology,and applied to the RoboCup simulation paltform.After repeated study,experiments show that interception technique have obtained desired result.
出处
《微计算机信息》
2012年第9期469-470,251,共3页
Control & Automation
关键词
激励学习
截球
多智能体系统
reinforcement learning
interception
multi-Agent system
