SVD-Aided EKF for Nanosatellite Attitude Estimation Based on Kinematic and Dynamic Relations

Abstract

Small satellite attitude angles are estimated using measurements of star trackers and rate gyros in this study. The issue related to gyro drifts is overcome by adding the bias terms into the state vector in order to estimate them. As an estimation method, two-stage non-traditional filter is used. In the first stage, singular value decomposition (SVD) is used for determining the attitude measurements. As a second stage, an extended Kalman filter (EKF) is designed based on linear attitude measurements. These two stages are integrated for the whole estimation algorithm in order to have estimations with high accuracy, and it is called SVD-Aided EKF. The proposed SVD-Aided EKF is used with two attitude models of satellite: only the kinematics model which does not include the dynamics of a satellite, and both kinematics and dynamics relations. Several scales of uncertainties on the principal moment of inertia of the satellite are considered in order to determine the level when estimation error of the kinematics and dynamics-based filter exceeds the error of the case using only kinematics relations.