Design and Flight Test Validation of a UAS Lateral-directional Model Predictive Controller

Abstract

Recent advances in computer technologies have increased the processing power on-board unmanned aerial aircraft. Computationally potent avionic systems have provided new opportunities to implement more adaptive and capable flight controllers. Model predictive control is emerging as a method for controlling unmanned aircraft, satisfying state and control constraints, and improving aircraft performance in the presence of external disturbances and nonlinear and unsteady aerodynamics. Although model predictive controllers provide many advantages over classical or modern control methods (such as PID or LQR), their practical applications have been limited to high-level path plannings, guidance logic, and control of slow robots with less complex dynamics. This work presents the development of an inner-loop model predictive control flight controller and successful validation and verification of its performance in actual flight tests. The work also investigates the impact of number of horizon points on the performance of the model predictive controller in the presence of wind and other external disturbances.