| This Thesis addresses problems in the dynamical modeling, identification, and control of underwater vehicles. First we report the design, construction, and testing by the Author of the Johns Hopkins University Remotely Operated underwater robotic Vehicle (JHUROV), a experimental test platform for research in underwater robotic vehicle navigation, dynamics, and control. Second we report a complete derivation of a general, six-degree-of-freedom finite-dimensional approximate dynamical plant model for rigid body underwater robotic vehicles, and an empirically justified simplification resulting in a single-degree-of-freedom finite-dimensional approximate dynamical plant model. Third, we report the novel use of and a stability proof for a scalar adaptive identification technique for identifying plant model parameters from experimental vehicle dynamics data. This technique is shown to produce superior experimentally identified dynamical plant model parameters in comparison to those obtained via a conventional least-squares technique. Fourth, we report a detailed comparative experimental evaluation of a family of model-based controllers designed for the task of trajectory tracking. Numerical simulation and extensive experimental results are reported, examining the comparative performance of five model-based controllers and a conventional linear controller. |
