Line-Of-Sight-based maneuvering control design, implementation, and experimental testing for the model ship C/S Enterprise I.

This report documents the progress, methods and engineering in building and testing the framework for CyberShip Enterprise 1. The main focus have been to modularize, standardize and improve the infrastructure and the operative system with respect to performance. With a secondary objective to create a user-manual to operate it. The work is done for a surface vessel in 3 degrees of freedom, surge, sway and yaw, and in calm waters and slow speed. The reliability when conducting laboratory experiments with CyberShip Enterprise 1 have greatly improved. This was achieved through repositioning and replacing the antenna for wireless communication, and a restructuring of how signals were handled between servers and softwares. The original setup with single frequency was replaced with a multifrequency producer-consumer structure. The futherst away a vessel is visible for Qualinsys is around 17 meters from the cameras. The shortest distance is roughly 4 meters infront of the cameras. This result in a lengthwise workspace for a vessel to be in the range of 13 meters. This length length can be increased to 27 meters, If the carriage start from the beach end, and moves toward the wavemaker during the experiment. It is not possible to increase it futher due to the length of the basin, the wavemaker, beach, in/outlet and carriage is occupying. In the system identication, an addtitonal hydrodynamic damping coecent for slow speed have been determined for the hull (Nv = 0:18140). In addtion higher order coecent have also been estimated from the towing measurements. A pseudo library for lookup table thruster mapping have been created. The bow thruster have been mapped for power limit input = f0.15, 0.3, 0.4, 0.5g, and the voith schneider propellers for speed input = f0.3, 0.4g. Measurments for voith schneider propellers for speed = 0.2 have also been conducted, but no lookup table have been created for it. The starboard voith schneider propeller rotates slow than the port voith schneider propeller. It is also noted that the servos are signicantly coupled and the force output drops at the periphery value , 1 . Of the servos, servo 4 have the strongest coupling in surge-sway. Two advance control design were implemeted, tuned and tested, a LgV backstepping and a Nonlinear PID designed controller. In the ideal simulated word both of them were equal in terms of maneuvering. Only ellipse path was used in the laboratory. This had to do with space constraints, and a wish to have long run time on the experiments. Originally only the LgV backstepping was tested in the laboratory. It converged and performed well. The only downside was it would constantly overshoot the heading, resulting iii iv Preface in a constant oscillation, while moving alone the path. The reason was due to the noisy velocity estimation. In the laboratory, their performance depended heavily on how they were implemented. If both of them were fully implemented, LgV backstepping proved to handle the uncertainties better. Although it would overshoot when exiting the sharper part of the ellipse path. While the fully implemented Nonlinear PID would struggle to converge to the path and behave erratic.The velocity dependent term distorted Nonlinear PID 's results the most. The Nonlinear PID is the superior one if only the rst term iss active. It would converge naturally to the desired position. Once on, it would stay there indenitely despite of the broken servo arm. With all the improvement in the reliability, one important problem still remains; the loss of visibility. There are incidents when Qualinsys displays it sees all markes, but it is unable to calculate the vessels position and orientation. The cause of this needs to be further investigated. Spare part servo arms should be purchased, and padding for the hull to dampen the impact to the basin walls. Several crack have been observed on the hull. Simulation Interface Toolkit have been discontinued, modifying it for Modular Interface Toolkit can be an option. The labeling and choice of variable name can be improved upon.