System identification and navigation of an underactuated underwater vehicle based on LSTM

collection: publications System identification is crucial for the pre-design, simulation, and control of underwater vehicles. However, typical identification methods usually require heavy equipment and incompetence of unmodeled dynamics. Therefore, we proposed a convenient technique for identifying both the hydrodynamic parameter model and the non-parameter dynamic model of underwater vehicle. After developing a fully coupled six degrees of freedom nonlinear model, we demonstrate a model-based EKF method for identifying all hydrodynamic damping coefficients without the need for specialized equipment. Additionally, a Long Short-Term Memory neural network is implemented to predict all linear and angular vehicle velocities, enabling navigation without the need for localization sensors.