Nonlinear Model Predictive Control for Agile Guidance of Fixed-Wing sUAS

We present an onboard receding-horizon optimization-based nonlinear guidance controller for path-following control of a fixed-wing small uninhabited aircraft in uncertain flow fields. We formulate two instances of the path-following optimal control problem as Model Predictive Contouring Control problems, and solve them in real-time using the ACADOS toolkit. We compare their performance to an industry-standard lookahead guidance method with hardware experiments. Our formulations lend themselves nicely to practitioners, as system identification of the employed kinematic model is straightforward. We demonstrate that our controllers are able to utilize the full control envelope, resulting in accurate path following. Finally, we utilize a state-of-the-art avionics stack and custom autopilot firmware to achieve real-time onboard solutions to the optimal control problems. Experimental test flights demonstrate the feasibility and effectiveness of the proposed approach on a fielded fixed-wing aircraft system.