Design of a Nonlinear State Estimator for Navigation of Autonomous Aerial Vehicles

The aim of this paper is to present a novel approach for the design and implementation of an onboard nonlinear state estimation system for an autonomous aerial vehicle. The tasks of such a system include collection of measurement data from different navigation sensors, and estimation of all the quantities describing the system’s state of motion based on the system dynamics and the measurement data. The widely accepted method for such a sensor fusion problem is the Kalman filter in the case of a linear system. The dynamics of a rigid body inherently involves nonlinearity, therefore a nonlinear extension of the Kalman filter is needed. The proposed solution relies on the Unscented Kalman Filter (UKF) technique with making use of both the quaternion and Rodrigues parameters representations of the attitude. The developed method is tested in MATLAB and implemented in the onboard embedded system in C code. The applicability of the proposed method is demonstrated in this paper by means of various examples.