Robust Control Architecture for Wind Rejection in Quadrotors

Abstract

Current efforts at the Advanced Dynamics and Control Laboratory (ADCL) at Embry-Riddle Aeronautical University (ERAU) are focusing on the implementation of robust control laws for disturbance rejection in quadrotors. This paper describes the development of two types of control architectures in an effort to reject or minimize wind effects in quadrotor UAVs. The design of a novel extension of the classic Non-Linear Dynamic Inversion (NLDI) control architecture for wind disturbance rejection is presented. This is followed by the application of adaptive artificial neural networks (ANN) to augment the classic NLDI control law designed to correct inversion errors caused by wind disturbance. Models are presented along with a simulation environment for various wind generated forces and moments. Monte Carlo numerical simulations are performed to analyze the performance of the classic NLDI, extended NLDI and NLDI with ANN augmentation under wind conditions. Results show that the NLDI with ANN augmentation outperforms the classic and extended NLDI controllers.