Stochastic feedback controller for a quadrotor UAV with dual modified extended Kalman filter

Abstract

In this paper, a filtering algorithm is proposed in order to improve the linearization procedure of the extended Kalman filtering (EKF). Our proposal consists of a parallel computing scheme, here called dual modified EKF (DMEKF), which comprises two algorithms to generate state estimates. One of the algorithms, namely Algorithm I, is a modification of the EKF, i.e. it differs from the EKF in that the real-time linear Taylor approximation is not taken at the previous estimate; instead, it is taken at the estimate by a second EKF algorithm, namely Algorithm II. Simulation results show that our proposal outperforms the EKF when trajectory tracking tasks are carried out by a quadrotor unmanned aerial vehicle (UAV) in a stochastic environment.