Nonlinear System Identification and Motion Control Design for an Unmanned Underwater Vehicle

This paper presents a hierarchical motion control system design for a fully actuated observation class Unmanned Underwater Vehicle (UUV). A Nonlinear Model Predictive Controller (NMPC), which uses the position measurement from an onboard Short Base Line (SBL) acoustical positioning system, is designed to control vehicle. The NMPC uses the measurement model to optimize vehicle trajectories with practical velocity and transponder visibility constraints to generate a reference velocity vector. The dynamics of the underwater vehicle is identified by using a polynomial representation of Nonlinear Auto-Regressive Moving Average with eXogenous input (NARMAX) model and using an onboard Doppler velocity log (DVL) sensor. The structure of the identified model is determined via Orthogonal Forward Regression (OFR) algorithm. The model parameters are selected based on the Error Reduction Ratio (ERR) criterion and are updated using a Recursive Extended Least Squares (RELS) algorithm. To track the reference velocities generated by the kinematics controller, a Control Lyapunov Function (CLF) based inverse optimal control law is synthesized. The proposed control scheme is validated both numerically and experimentally showing that dynamics of vehicle can be captured successfully by the proposed method. It is also shown that the proposed controller can effectively track the desired references.