For a class of multi-input multi-output(MIMO) block nonlinear systems with mismatched uncertainties, an
adaptive controller design scheme using backstepping and RBF neural networks is proposed. By introducing a modified
Lyapunov function, control singularity problem brought by unknown control matrices is avoided. By using of robust adaptive
control technique, many difficulties brought by MIMO uncertainties are solved. The conclusion is obtained that all states
variables are bounded and will exponentially converge to a neighborhood of the origin globally. Finally, simulation results
are given to show the correctness of proposed scheme.