Mapping　is　an　important　research　direction　in　the　field　of　robot　navigation.　Map　represents　the　robot＇s　working　environment　has　very　important　effect　on　localization　and　path　planning.　Vision　sensor　is　simple　and　reliable,　and　has　been　widely　used　in　the　field　of　mapping.　The　paper　introduces　the　process　in　which　the　binocular　stereo　vision　system　is　adapted　to　imitate　the　double　eyes　of　mankind　in　order　to　realize　computer　vision.　The　autonomous　mapping　algorithm　in　this　paper　is　based　on　GSOM　(Growing　Self-organizing　Map)　algorithm.　By　continuously　increasing　new　neurons　in　the　growth　of　network　scale,　the　GSOM　mapping　algorithm　applies　a　small　number　of　SOM　neurons　to　generate　a　topologic　map　describing　the　environmental　characteristics.　The　experimental　results　show　that:　we　applied　GSOM　algorithm　to　autonomously　generate　a　topological　map,　which　was　accurately　described　by　a　small　number　of　SOM　neurons.　©　2010　IEEE.