Multi-Mode Extended Kalman Filter for Aircraft Attitude Estimation

Abstract

AbstractThis paper introduces a new multi-mode Extended Kalman Filter (EKF) algorithm for attitude estimation of small UAVs during the whole flight (from take off to landing). At first, it examines the available INS/GPS measurements from the point of view of applicability in the estimator on ground and in air. From this, a multi-mode EKF is developed which switches between measurements to use them optimally. The quaternion representation of rotation was used to avoid singularity and derive a closed form solution of the Heun scheme in the discretization of system dynamics. Filter initialization and observability issues are also covered. After presenting the computational steps of the EKF the hybrid automata representation is described. This includes the description of estimator modes, automata states and input events. Finally, the graph of the automata representation is published. The paper ends with the description of tuning process and presentation of off-line test results on real noisy data. The new EKF performed well during all the tests including flights with stabilization controllers based on the estimated Euler angles.