Efficient Extended Kalman Filtering for attitude estimation based on gyro and vector observations

摘要

This paper presents an efficient Extended Kalman Filter (EKF) which can simultaneously process all the observation information. The EKF applies the Rodrigues parameters of attitude and gyro drift-rate bias vector as the state vector. A new third order measurement equation is used which contains all the observation information. By using this measurement equation, the computation of the gain matrix needs to take the inverse of a 3×3 matrix only once. The measurement sensitivity matrix is not singular if the number of observed vector is not 1. Simulation results show that the proposed EKF improved the computational efficiency without accuracy loss.