A new method is proposed for the fault detection and identification of wheeled mobile robots with uncertainties and disturbances based on the mathematical model of a multivariable input affine nonlinear dynamic system; The extended parity space method is used to eliminate the unknown variables of the mobile robot system, so as to obtain the nonlinear analytical redundant residual for the fault detection of the mobile robot, and the corresponding fault feature matrix is given according to the structure form of the residual in the fault function for the fault identification of the mobile robot; The experimental results of Shihe MR500 mobile robot show that the method is effective.