To　deal　effectively　with　uncertainty　involved　in　3D　map　building　for　mobile　robot　and　enhance　the　accuracy,　completeness　and　consistency　of　the　map,　this　paper　proposed　a　3D　simultaneous　localization　and　mapping　(SLAM)　approach　based　on　fusion　of　sensor　data　and　Rao-Blackwellised　particle　filtering　(RBPF).　The　sensor＇s　uncertain　probability　model　was　established　and　noise　involved　in　the　sensor＇s　data　is　dramatically　removed　by　Bayesian　filter.　The　sensor＇s　information　gathered　from　vision　sensor　and　laser　range　finder　was　integrated　by　a　sophisticated　rule,　and　the　3D　environment　map　with　texture　mapping　was　established　in　SLAM　framework.　As　demonstrated　from　experimental　results,　the　autonomous　SLAM　routine　based　on　sensor　fusion　is　qualified　for　building　an　elegant,　complete　and　accurate　environment　model,　which　verifies　the　effectiveness　of　the　presented　approach.　©,　2015,　Beijing　Institute　of　Technology.　All　right　reserved.