Analysis, experimental evaluation, and software upgrade for attitude estimation by the Shallow-Water AUV Navigation System (SANS)

Abstract

The main problem addressed by this research is the lack of a small, low-cost integrated navigation system to accurately determine the position of an Autonomous Underwater Vehicle (AUV) during all phases of an underwater search or mapping mission. The approach taken utilized an evolving prototype, called the Shallow-Water AUV Navigation System (SANS), combining Global Positioning System (GPS), Inertial Measurement Unit (IMU), water speed, and magnetic heading information using Kalman, low-pass, and complimentary filtering techniques. In previous work, addition of a math coprocessor improved system update rate from 7 to 18 Hz, but revealed input/output coordination weaknesses in the software. The central focus of this thesis is on testing and programming improvements which resulted in reliable integrated operations and an increased processing speed of 40 Hz. This now allows the filter to perform in real time. A standardized tilt table evaluation and calibration procedure for the navigation filter also was developed. The system was evaluated in dynamic tilt table experiments. Test results and qualitative error estimates using differential GPS suggest that submerged navigation with SANS for a period of several minutes will result in position estimation errors typically on the order of 10 meters rms, even in the presence of substantial ocean currents.