QP-MPC Based Guidance for Complex Path Following of Fixed-Wing Aerial Vehicles

Recent advances in autonomous aircraft guidance have produced a wide range of geometric, kinematic, and nonlinear control strategies, however, most modern approaches rely on fixed gains, simplified path parameterizations, or prior knowledge of the reference trajectory, limiting their ability to generalize to collision avoidance scenarios, nonlinear path following tasks, or disturbance-rich environments. Model Predictive Control (MPC) offers a powerful alternative by optimizing control inputs over a finite horizon and explicitly enforcing constraints on states, dynamics, and control effort. However, despite notable progress in real-time onboard MPC for autonomous aircraft, classical MPC formulations remain computationally intensive for payload- and power-constrained small fixed-wing aircraft with limited payload capacity, where size, weight, and energy restrictions severely limit onboard processing resources. This paper introduces a quadratic-program reformulation of model predictive control (QP-MPC) that enables real-time guidance for a payload-stringent small fixed-wing aircraft. By leveraging linearization along a receding trajectory manifold and converting the MPC problem into a condensed quadratic program, the proposed method achieves computational speeds comparable to widely used geometric guidance laws such as L1 approach, while retaining MPC’s inherent robustness and constraint-handling capabilities. Comprehensive simulation studies as well as realistic atmospheric disturbances generated with a Dryden turbulence model under high wind speeds demonstrate that the QP-MPC algorithm delivers superior path-tracking accuracy, improved resilience to severe disturbances, and enhanced generalization to unstructured collisionavoidance and nonlinear path-following tasks while significantly reducing the computational expense associated with classical MPC implementations.