Dual-layer optimization-based control allocation for a fixed-wing UAV

Many existing control allocation methods separate the high-level control design from their low-level allocation design, assuming that the constraints of actuators can be guaranteed by the allocator. This idea may not be suitable for the nonlinear fixed-wing unmanned aerial vehicle studied here, which hence motivates this work. In this paper, we propose a new dual-layer optimization-based control allocation method, in which the proposed allocator, on the one hand, can modify the pre-designed virtual signals from the high-level when the out-layer actuator, i.e., throttle, reaches its constraint. On the other hand, it reverts the conventional constrained allocator when the throttle constraints are inactive. Another feature is that under the proposed framework, the initial state of the augmented actuator dynamics serves as design parameters, bringing more degrees of freedom for allocation design without affecting the nominal stability. Apart from the control allocator, this paper also proposes a high-level flight controller based on the control-oriented model and a combination of nonlinear dynamic inversion and disturbance observer. Disturbance observer provides robustness by estimating the model errors between the control-oriented and true models, and compensating for them in the controller. High-fidelity simulation results under realistic wind disturbances are presented to demonstrate the performance of the proposed method.