Adaptive calibration of an autonomous underwater vehicle navigation system

Abstract

There continues to exist the problem of long-term accurate position estimation for autonomous underwater vehicles (AUVs). In current operations, the AUVs positional fix is initially obtained on the surface from a global positioning system (GPS) receiver. The AUV then submerges to perform the desired mission. While submerged, location/navigation is performed using, at a minimum, an inertial navigation system (INS). Depending on the sophistication of the AUV, Doppler velocity sonar (DVS) might be combined with a multi-state Kalman filter (KF) to perform position estimation. The estimates from the INS and the DVS/Kalman filter estimator are combined to provide a robust estimate of location. Because the KF is model based, there is a likelihood that over time the divergence of the KF may increase since the true motion of the AUV does not match the modeled motion. At that point the AUV must surface to obtain another set of absolute position coordinates from the GPS before being able to continue its mission. Depending on the duration of the mission, this process may need to be repeated several times, which unnecessarily uses battery/power resources. By combining a database which contains sonargrammetric, terrain matching, and image registration information, with the standard navigation instrument suite, the accuracy of positional estimates could be maintained over a longer duration. This would allow the AUV to remain submerged for longer periods of time, thus minimizing the drain on the limited power resources.