13动态环境下基于GNG网络拓扑地图创建_仲朝亮.pdfVIP

Vol. 38 No. 1 ( ) 63 2012-02 Journal of East China U niversity of Science an T echnology ( Natural Science E ition) : 1006-3080( 2012) 01-0063-06 GNG 1, 2 1, 2 1, 2 仲朝亮 , 刘士荣 , 杨 帆 ( 11 华东理工大学自动化研究所, 上海200237; 21 杭州电子科技大学自动化研究所, 杭州310018) : 动态障碍 的存在导致产生不一致的环境地图, 为此设计了一种新的拓扑地图创建方 法该方法首先通过概率方法过滤运动障碍 的干扰信息, 获得环境可行区域信息, 再将可行区域 信息作为GNG 算法的输入空间, 通过学习与不断增加新的拓扑节点, 创建一致的环境拓扑地图 该方法具有自学习自适应等特点通过仿真与 理实验验证了其可行性与有效性 : 动态环境; 地图创建; 拓扑网络; 移动机器人 : TP242 : A Topological Map Building Based on GNG Network in Dynamic Environment 1, 2 1, 2 1, 2 ZH ON G Chao-liang , L I US hi-r ong , YA N G F an ( 11 I nstitute of A utomation, East China Uni ersity of Science and Technology , Shanghai 200237, China; 21I nstitute of A utomation, H angz hou D ianz i Uni ersity , H angz hou 310018, China) Abstract: Dynamic obstacles may result in inconsistent environment map. A iming at this problem, a new buil ing metho of topological map is esigne an realize for mobile robot. Firstly , the feasible regioncs information is obtaine via a probabilistic metho to filtrate spurious measurements of ynamic obstacles. An then, w ith this information as the growing neural gas ( GNG) algorithmc s inputs, a consistent topological map is built by learning an a ing new no e. T his propose metho is self-learning an a aptive. Both the simulation an physical experiments verify the feasibility an effectiveness of the propose metho . Key wor