Navigation and target localization performance of the autonomous underwater vehicle REMUS

Abstract

A frequent stipulation in the design of Autonomous Underwater Vehicles (AUVs) is the requirement that the vehicle be small and inexpensive. Such a constraint precludes the use of costly, highly accurate sensors. As a result, to achieve a highly accurate and robust navigation system, navigation data from any and all sources must be processed. This information may come from a number of sources such as an Acoustic Doppler Current Profiler, Long Baseline acoustic travel times to known beacons, Inertial Measuring Units, or Sonar. The objective of this research was to develop a Kalman filter-based navigation algorithm for the AUV REMUS that improves positioning accuracy, provides rejection of poor fixes, and decreases energy use due to excessive corrections in course. Research was conducted in the context of Naval Special Warfare and its current vision for use of the REMUS vehicle in shallow water mine hunting. Navigation performance is illustrated using REMUS data for a Phase I search of a shallow water environment. Results are presented from a navigation sensor data fusion algorithm being developed for this scenario. Results demonstrate outlier rejection and track smoothing, both of which are beneficial to improving sensor data and increasing the reliability of target reacquisition.