Preliminary results in experimental identification of 3-DOF coupled dynamical plant for underwater vehicles

Underwater vehicle control research is presently limited by the availability of explicit experimentally validated plant models. We report an experimentally identified 3 degrees of freedom (DOF) coupled non-linear finite-dimensional plant model for the Johns Hopkins remotely operated vehicle (JHUROV) for the forward, lateral, and heading degrees of freedom. We report the parameter estimation methodologies of total least squares (TLS), ordinary least squares (OLS), and their underdetermined variants. We report that in general TLS is superior to OLS when error exists in all data. We report a 3-DOF coupled non-linear finite-dimensional plant of an underwater vehicle estimated by TLS and compared to the same plant identified by OLS by comparing the mean absolute error between the velocity profile of a numerical simulation and the experimental velocity. We report that the TLS estimate of the parameter model including a fully parameterized quadratic drag performed best overall in cross validation and that in general TLS estimate performs better then OLS estimate. Our goal is to enable the development of new model based controllers capable exact position and velocity trajectory tracking of the forward, lateral and heading DOFs.