To　obtain　the　information　of　unknown　environment　as　sufficient　and　fast　as　possible,　this　paper　proposed　an　effective　approach　to　build　a　local　topological　map　of　unknown　environment.　The　laser　data　obtained　by　laser　sensor　is　successfully　transformed　into　a　local　grid　map　refined　by　Bayesian　rule　and　subsequently　a　skeleton　of　the　free　area　in　the　generated　local　grid　map　is　extracted　via　an　improved　skeleton　algorithm　which　was　also　proposed　in　this　article.　In　addition,　a　set　of　topological　node　selection　rules　on　the　previous　extracted　skeleton　are　carefully　designed,　and　the　generated　node　group　is　organized　into　a　tree　data　structure.　Thus　the　local　topological　map　is　established　based　on　the　laser　data　in　unknown　environment.　As　demonstrated　from　experimental　results,　the　proposed　approach　is　effective　to　reveal　genuine　structure　of　the　robot＇s　environment.　©　2015　IEEE.