Identification of an Autonomous Underwater Vehicle Dynamic Using Extended Kalman Filter with ARMA Noise Model

In the procedure of designing an underwater vehicle or robot, its maneuverability and controllability must be simulated and tested, before the product is finalized for manufacturing. Since the hydrodynamic forces and moments highly affect the dynamic and maneuverability of the system, they must be estimated with a reasonable accuracy. In this study, hydrodynamic coefficients of an autonomous underwater vehicle (AUV) are identified using velocity and displacement measurements, and implementing an Extended Kalman Filter (EKF) estimator. The hydrodynamic coefficients are included in the augmented state vector of a six DOF nonlinear model. The accuracy and the speed of the convergence of the algorithm are improved by selecting a proper covariance matrix using the ARMA process model. This algorithm is used to estimate the hydrodynamic coefficients of two different sample AUVs: NPS AUV II and ISIMI. The comparison of the outputs of the identified models and the outputs of the real simulated models confirms the accuracy of the identification algorithm. This identification method can be used as an efficient tool for evaluating the hydrodynamic coefficients of underwater vehicles (robots), using the experimental data obtained from the test runs. 1. Introduction In recent years, an extensive researches has been conducted in the area of underwater robotics and underwater vehicles. Advanced estimation and control methods have been used in order to improve the capability of AUV positioning and path tracking. In [1], an experimental method is implemented for a mobile underwater vehicle to determine its hydrodynamic coefficients. In [2], EKF is used for localization and mapping of autonomous mobile robots. The hydrodynamic coefficients of an AUV were identified in [3] using an EKF. A six degree-of-freedom model of motion was developed in [4] for an underwater vehicle, where an autopilot system was designed for automatic sliding mode control of the vehicle. To examine the maneuverability and performance of the control system of an AUV, a mathematical model of the vehicle must be identified. The mathematical model includes the hydrodynamic forces and moments, expressed in terms of hydrodynamic coefficients. Therefore, estimating the exact values of these coefficients is an important step in modeling of an AUV. It has been observed that the linear