Adaptive Flight Control Design for Quadrotor UAV Based on Dynamic Inversion and Neural Networks

An oblique cross quad rotor UAV is designed and assembled. To achieve autonomous flight control, a nonlinear dynamic model is investigated. Then an adaptive flight control theory, based on dynamic inversion and linear neural network, is introduced to the control of the UAV. Based on the time-scale separation principle, an attitude dynamic inverse controller and a trajectory dynamic inverse controller are deduced respectively according to the nominal nonlinear model. The inverse error dynamics is regulated with a PD controller. A Sigma-Pi neural network is introduced to eliminate the inverse error adaptively to improve the robustness of the controller. To verify the flight controller, two group simulations are conducted: the first is to validate the robustness of the attitude controller under severe parameter uncertainty. Then a compound MTE (mission-task-elements) simulation, which includes pirouette and vertical maneuver, is applied to the verification of the flight control system. The simulation results including an imitation of inverse error and disturbance moments are provided. The demonstrations show that the designed flight control system has adaptability and robustness and can achieve accurate attitude and trajectory control.