基于自组织与LVQ神经网络的足球机器人协作策略学习
Coordination strategies learning of soccer robot based on self-organizing and LVQ NN
-
摘要: 足球机器人系统目前已成为人工智能应用技术研究的重要实验平台,系统的核心部分就是决策子系统.本文主要研究机器人足球比赛中协作策略的学习问题,采用了自组织与学习向量量化(LVQ)神经网络算法实现了两个足球机器人的传球学习.提出了一种协作策略学习的改进算法.该算法通过对网络权值矩阵进行改进,可以显著提高网络的训练质量.仿真和实验结果表明了该方法的可行性与有效性.Abstract: Soccer-robot system now has been an important experiment platform for artificial intelligence application technology.The key point of the system is the decision-making subsystem.This paper focuses on coordination strategies learning in robot soccer,self-organizing and learning vector quantization neural networks.A new improved method of coordination strategies is presented in this paper.It is obvious to improve the quality of network training by reforming the NN weight-matrix.The result of simulation experiment indicates that the proposed method is suitable and effective.