A method is proposed to solve the inverse kinematics and control problems of robot control systems using a cerebellar model articulation controller neural network combined with a genetic algorithm. Computer simulations and experiments with a 7-DOF redundant modular manipulator have demonstrated the effectiveness of the proposed method.