This paper proposes an Extreme Learning Machine (ELM), robust controller for unmanned aerial vehicle (UAV) quadrotor. The proposed method is applied to the inner and outer loops for position and attitude control. A single hidden layer feedforward network based on ELM whose input weights and hidden node parameters are generated randomly and fixed, the output weights are updated using optimization method, by this way we guarantee the fast learning speed to drive the system to the desired trajectory. Precise dynamic model and prior information of disturbances are not needed. The simulation study is presented to show the effectiveness of the proposed control algorithm comparing it with the conventional neural network