The　most　characteristic　of　Lunar　Rover　Motion　Planning　and　Control　is　unstructured　of　lunar　terrain,　and　do　not　establish　accurate　mathematical　model.　In　order　to　used　the　method　of　environmental　model　and　analysis　to　study　the　lunar　rover　motion　planning.　Combining　the　nature　of　convex　combination　on　this　paper,　proposes　the　fuzzy　neural　network　system　based　on　SAM　which　apply　particle　filter　training　algorithm.　Proved　the　SAM-FNN　is　continuity,　stability　and　accessibility.　The　Lunar　Rover＇s　translational　speed　and　rotation　speed　are　smooth　and　continuous　changes.　Particle　filter　training　algorithm　to　overcome　the　weakness　that　current　training　algorithms　of　Neural　Network　is　likely　to　trap　in　local　minimum.　It　is　an　efficient　dealing　with　nonlinear/non-Gaussian　problems.　Simulation　results　show　that　its　performance　is　markedly　superior　to　those　available.