Constrained Filtering Method for MAV Attitude Determination

Abstract

In this paper we introduce a novel attitude measurement algorithm that gives the Euler angles based on MEMS sensors including three rate gyros, three magnetometers and one accelerometer. Six elements of the direction cosine matrix are selected to compose a state vector. Only four elements are independent, because there exists two nonlinear constraints. They are introduced as virtual measurements in the Kalman filter. The filter can be improved by introducing heading differential speed signal. Bias of simulation results indicates that the Euler angles can be determined with standard deviations at 1-3 degree even during high dynamic maneuvers and long turns. Flight tests have been carried out in a miniature UAV, the Euler angles' in which the method is proved to be valid in high dynamic maneuver