A tracking of AUV with integration of SSBL acoustic positioning and transmitted INS data

Abstract

An autonomous underwater vehicle (AUV) can achieve precise observation in the sea. Also in Japan agency for Marine-earth Science and Technology (JAMSTEC) some observations with AUV ldquoURASHIMArdquo have been implemented mainly in deep ocean. Tracking or navigation of the AUV is significant to make the observations more valuable. Almost AUV equips an inertial navigation system (INS) to navigate own cruise. The main problem of INS is the drift error of the estimated position, velocity and attitude, and it is not negligible. On the other hand, the AUV is tracked or positioned using acoustic positioning system (APS) in many cases. The APS is direct estimation of the position, and the error is limited but very larger compared with short-time INS tracking. In JAMSTEC, the super short baseline (SSBL) method is almost used to track the AUV from the mother ship because the observation area is deepwater and the SSBL is easy to operate. However, the SSBL is not accurate particularly in deepwater. Furthermore, there is systematic error due to measurement error of the circumstance, such as sound velocity profile. Thus the SSBL estimation is not so good to use to reset the INS estimation of the AUV position. In this paper, an accurate tracking method to estimate AUV potion by SSBL from the mother ship is proposed. In this method, the INS data obtained in the AUV is transmitted to the ship by spectrum spread (SS) acoustic communication. This acoustic signal is also used for SSBL, and the SSBL estimation is implemented by each data symbol. Thus the SSBL estimation is achieved not with a short pulse but with a continuous acoustic signal. The SSBL estimation results and the transmitted INS data are integrated and the AUV position is tracked accurately. Two integration algorithms are introduced. The one considers only the drift error of the INS outputs. This algorithm is linearly modeled and filtered by a Kalman filter (KF). Another also considers the systematic error of the SSBL due to measurement error of layered sound velocity profile. The later algorithm is non-linear and uses extended KF (EKF). The usefulness of these algorithms is shown by simulation. Particularly, the second algorithm is beneficial under existence of the systematic error of the SSBL estimation.