This　paper　addresses　the　issues　of　the　autonomous　exploration　and　grid-topology　hybrid　map　building　of　mobile　robots　in　large-scale　environments.　First,　in　order　to　improve　the　defect　of　the　original　method　in　the　special　environment,　a　way　to　extract　the　target　point　based　on　the　free　area　is　proposed.　Then,　we　use　the　information　gain　and　the　extraction　method　of　fast　back　to　the　initial　topological　node　to　evaluate　the　target　point,　so　as　to　solve　the　problem　that　the　long　time　no　loop　closure　detection　leads　to　the　failure　of　the　map　building.　Finally,　we　propose　an　improved　method　of　building　and　optimizing　the　real-time　topological　map　based　on　the　information　of　the　laser　data　and　grid　map,　which　makes　the　grid-topology　hybrid　map　more　concise　and　the　exploration　process　more　efficient.　The　effectiveness　of　the　proposed　autonomous　exploration　and　grid-topology　hybrid　map　building　method　for　mobile　robots　in　unknown　environments　is　verified　by　experiments.