Environmental　map　building　is　an　important　premise　of　robot　navigation,　Self-Organizing　Map　(SOM)　or　Growing　Self-Organizing　Map　(GSOM)　can　establish　the　topological　map　of　the　indoor　environment　through　the　environmental　cognition.　Aimed　at　the　problem　of　traditional　SOM　containing　error　structure　in　constructing　topological　map　while　long　time　consuming　of　GSOM,　this　paper　proposes　a　Prunable　Self-Organizing　Map　(PSOM)　algorithm　which　can　prune　wrong　connections　between　neurons　from　the　environmental　structure,　thus　proving　accurate　topology　while　greatly　reducing　the　implementation　cost　of　the　program　compared　to　the　Growing　Self-Organizing　Map.　The　proposed　method　(PSOM)　is　verified　by　experimental　of　environment　map　building　based　on　physical　indoor　environment.　©　2018,　Springer　Nature　Singapore　Pte　Ltd.