Application of Extended Kalman Filtering to a Model-Based, Short-Range Navigatorfor an AUV

摘要

Autonomous Underwater Vehicles (AUV) are being considered by the Navy forperforming a variety of missions. During the research and development state of the AUV project a navigator is needed to provide vehicle position estimates for shortrange missions performed in a test pool environment. This navigator should operate with inexpensive sensors and not require excessive digital processor time. This thesis presents the results of the design of a model-based navigator. The navigator uses nonlinear vehicle models of Extended Kalman filter theory. Simulation studies for both a 12,000 pound vehicle and the 435 pound testbed vehicle, designed and built at the School (NPS AUV II), are presented. Results of using data recorded from the gyroscopes and depth cell installed in the NPS AUV II vehicle in lieu of simulated data are also discussed. These results show that the navigator meets the goals of low cost and low processor burden for short-range missions.