Nonlinear H∞ Control based on Gaussian Process for a quadrotor

Abstract

The use of Unmanned Aerial Vehicles (UAVs) can be subject to external disturbances, such as wind gusts, aerodynamic forces, and parametric uncertainties, depending on the flying environment. Therefore, the automatic control must be robust enough to execute a predefined task. In this work, we propose a robust control strategy based on the dynamic model to mitigate external disturbances and parametric uncertainties. The H∞ criterion is employed to provide robustness and Gaussian processes are used to supplement the mathematical model of the quadrotor, estimating uncertainties not accounted by the nominal model. A comparative study among the proposed controller, the nonlinear H∞ controller, and the neural network-based nonlinear H∞ controller is presented, in which we analyze trajectory tracking under different levels of parametric uncertainty and external disturbances, and compare the performance and energy consumption of the controllers.