Cubature Kalman Filtering Fusing Expectationmaximization Method for Attitude Determination Applied to Underwater Glider

The inertial navigation system (INS) aided with the magnetic compass (MCP) is designed for the underwater glider to achieve higher accuracy, higher stability and faster tracking even if sudden change of attitudes. The CKF has the better estimation accuracy even though measurements become unobservable, but the computational load of CKF is much higher than EKF. The paper presents a novel method fusing CKF and Expectation-maximization (EM) (CKFEM) which makes fully use of the anti-interference advantage and the computational efficiency to acquire high accuracy navigation information for a long period of gliding. In the real glider experiments, the CKFEM has a good effect in the aspects of accuracy, stability and tracking even if the sudden change of attitude.