The present research paper deals with identification of an Autonomous Underwater Vehicle (AUV) system using Kalman filter and maximum likelihood estimation approaches in order to get an accuracy and more representative nominal model. Despite all efforts in modelling of the vehicle, uncertainties in the parameters can appear because some dynamics are difficult to be modelled and, eventually, are neglected. Complementary, the paper aims the problem of robust control in the sway and yaw dynamics assuming uncertainties, perturbations and noises. The results are shown by simulated data. © 2011 IFAC.