|dc.description.abstract||The objective of this thesis is to investigate the modeling and control of modular underwater robots. This objective is motivated by recent events in the offshore industry, where innovative solutions are needed to cope with the upcoming challenges. The vision is to use small-sized modular underwater robots to inspect all areas of offshore structures efficiently, but at the same time maintain the capacity for intervention through morphological changes in the system.
This thesis concerns itself with modeling and docking control of a system composed of modular underwater robots. The first part of the thesis consists of five chapters, which combined investigates the mathematical modeling of a system of modular underwater robots with arbitrary interconnection between them. The first chapter presents the kinematics, the marine vehicle models, and the notation. An essential feature of a Modular Underwater Robot (MUR) system is the ability to reconfigure the morphology of the interconnection. Thereby, the underlying modeling methodology must handle structural changes in the system. The second chapter of the first part introduces different modeling approaches with examples before presenting the chosen modeling approach. Furthermore, the chapter develops a model for the MUR system into a simulator and verifies the implementation by a numerical investigation.
Any dynamic model suffers from imperfect model knowledge, and aggregating multiple models into a large-scale model only magnifies the effect. The subsequent chapter of Part I validates the developed modeling approach by subjecting multibody systems to different experiments. The chapters compare the behavior of the real and the simulated system, respectively, and seeks to quantify the concordance between them.
The automatic modeling method, developed in the first part of the thesis, applies when MURs gather into a morphology. The second part of this thesis concerns with the MUR system before the aggregation of the MURs. The aggregation of the MURs require them to approach each other, called Rendezvous, and then, physically connect to each other, called docking. The considered rendezvous and docking problem is assumed to be camera-based, such that, the navigation of the systems utilize cameras for position estimation. The camera introduces line-of-sight conditions that must be kept. Part II proposes to employ distributed predictive control for solving the camera-based rendezvous and docking problem. The predictive controller is capable of embedding the line-of-sight constraint directly in the formulation, while synchronizing the rendezvous pose between the vehicles.||nb_NO