Kalman filter estimation of underwater vehicle position and attitude using Doppler velocity aided inertial motion unit

Abstract

This Paper explores the use of an extended Kalman filter to provide real-time estimates of underwater vehicle position and attitude. The types of previously available sensors are detailed including strapdown accelerometers, roll and pitch sensors, gyro and magnetic compasses, depth sensor, and various types of acoustic positioning systems. A doppler velocimeter is added to this sensor suite to improve the performance of the filter. As an integral part of the filter, magnetic compass and gyrocompass biases are estimated to improve vehicle heading accuracy. The filter is designed to account for numerous reallife complications. These include varying rates of sensor output, lengthy gaps in reception of position information, presence of non-Gaussian position fix errors (flyers), and varying probability density functions for sensor errors. Simulated data are used to test the filter with varying availability of data and accuracy of initial conditions, along with actual data from a deployment of the towed DSL-120 vehicle. The increased accuracy obtained by using the doppler velocimeter is emphasized.