For　the　collaborative　formation　control　problem　of　a　multi-snake　robot　with　uncertain　sideslip　and　lumped　unknowns,　this　study　proposed　collaborative　line-of-sight　(LOS)　guidance　and　consensus-based　path　variable,　which　enables　snake　robots　to　track　multiple　parametric　paths　in　a　fixed　formation.　Meanwhile,　the　improved　finite-time　sideslip　observer　effectively　compensates　for　the　kinematic　deviation,　thereby　improving　the　accuracy　of　the　multi-snake　robot＇s　path　tracking.　A　robust　adaptive　controller　based　on　an　integral　sliding　mode　manifold　is　developed　to　eliminate　external　disturbances.　The　model　uncertainty　and　time-varying　disturbances　are　comprehensively　involved　and　incorporated　as　lumped　unknowns　in　dynamics.　The　prediction-based　RBF　neural　network　estimators　are　derived　to　approximate　unknowns　and　compensate　controllers.　The　Lyapunov　theory　is　adopted　to　prove　the　stability　of　formation　errors　for　snake　robots　in　the　closed-loop　system.　Simulation　and　experiment　results　show　the　practicability　and　superiority　of　this　work.　©　2004-2012　IEEE.