Attitude Estimation Fusing Quasi-Newton and Cubature Kalman Filtering for Inertial Navigation System Aided With Magnetic Sensors

In the complex underwater environment, the performance of microelectro-mechanical system sensors is degraded sharply and the errors will become much larger. Especially when the magnetic sensor is disturbed by the external magnetic interference, the measurements become unobservable so that the navigation information is estimated erroneously. To solve this problem, the paper proposes a novel method fusing Quasi-Newton and cubature Kalman filter (QNCKF). This method takes full advantage of the computation efficiency of the Quasi-Newton and the estimation accuracy of CKF in the case of nonlinearity. The performance of QNCKF is verified theoretically and evaluated by experiments. The results indicate that when the magnetic sensor is interfered, QNCKF and CKF still can maintain high estimation accuracy, whereas the extended Kalman filter performs poorly. Moreover, QNCKF is superior to CKF in the aspect of computational efficiency. Therefore, QNCKF has the highest priority in terms of estimation accuracy and computational efficiency among the three methods and it is more suitable to be applied to the underwater gliders than the other two methods.