A　method　for　mobile　robots　human　tracking　based　on　the　double-layer　cooperative　locating　mechanism　is　proposed　to　realize　the　tracking　in　complex　environments.　The　proposed　method　implements　the　coarse　location　and　the　fine　location　based　on　the　data　from　radio　frequency　identification　(RFID)　and　stereo　vision　to　improve　the　locating　accuracy.　First,　the　coarse　position　of　the　given　person　is　estimated　by　RFID.　Then,　the　processing　techniques　of　extended　Kalman　filter　(EKF)　and　the　adaptive　template　matching　algorithm　are　applied　to　the　narrowed　ROI　images　for　precise　position　based　on　the　feature　of　head-shoulder　and　motion.　Finally,　the　intelligent　gear　shift　control　strategy　considering　the　distance　between　human　and　robot　is　utilized　to　drive　the　robot　towards　the　given　target.　The　experimental　results　show　that　the　presented　method　can　deal　with　the　case　when　there　is　an　occlusion　or　a　sudden　turning.