A　composite　adaptive　control　of　free-floating　dual-arm　sptial　robot　system　with　an　uncontrolled　base　was　proposed.　In　order　to　deal　with　the　problem　that　the　dynamics　equations　have　a　nonlinear　relationship　with　the　initial　parameters,　the　spatial　robot　can　be　modeled　as　an　under-actuated　system.　And　it　is　shown　that　the　dynamics　equations　of　the　system　can　be　linearly　parameterized.　With　the　augmentation　approach,　we　demonstrate　that　the　augmented　generalized　Jacobi　matrix　can　also　be　linearly　parameterized.　Based　on　the　results,　the　composite　adaptive　control　scheme　was　designed　to　track　the　desired　trajectory　in　workspace　for　dual-arm　spatial　robot　with　unknown　parameters.　The　experimental　results　validate　the　proposed　method.