A　tactile　sensor　is　a　necessary　means　for　intelligent　equipment　to　acquire　external　environment　information　and　improve　the　performance　of　human-robot　interaction.　Although　high-performance　tactile　sensor　array　is　widely　studied,　the　large　number　of　wires　required　to　transmit　data　from　numerous　arrays　is　still　a　major　obstacle　in　large-area　application.　In　this　study,　a　large-area,　low-cost,　stretchable,　textile-based　tactile　sensor,　which　is　sensitive　for　contact　position,　is　proposed.　The　sensor　has　a　simple　three-layer　structure　and　four　external　wires　due　to　the　use　of　a　novel　double-faced　effect　functional　knitted　textile　with　＂self-uniformity＂　characteristic.　The　porous　polyurethane　foam　with　a　large　pore　size　is　first　reported　as　touch　switch　material.　It　not　only　has　excellent　touch　switch　function,　but　also　makes　the　sensor　have　a　soft　and　elastic　touch　and　good　impact　buffering.　In　addition,　the　application　of　radial　basis　function　neural　network　makes　the　sensor　have　self-learning　＂intelligence,＂　which　makes　the　sensor　flexibly　and　quickly　arrange　even　on　the　surface　of　a　complex　3D　object.　Finally,　the　potential　applications　of　the　sensor　are　demonstrated.　This　study　shows　that　the　sensor　has　great　potential　in　the　fields　of　wearable　devices,　robot　interaction　control,　and　human-computer　interface.