Design And Evaluation Of Backstepping And Model Predictive Controllers For An Over-actuated Marine Platform.

Abstract

Robotics is a technological science that deals with the design, construction, operation,and application of robots, as well as computer systems for their control, sensory feedback,and information processing. The calculation of the signals, which control the actuators,is performed by control algorithms. In several cases, the application of novel controlmethods to robotic systems, such as backstepping and model predictive control, has ledto a deeper understanding of their mechanical properties.In the diploma thesis, we focus on a particular class of robotic systems, the over-actuatedrobotic systems. We present the design of a backstepping controller and a model predic-tive controller for a triangular floating platform, aiming at the stabilization of its linearand angular velocities as well as its position and orientation. Three rotating jets, locatedat the corners of the platform, control its motion. With this control configuration, theplatform is over-actuated.Some significant points are the following. In the design of the backstepping controller,we have taken into consideration the delay in the application of the forces and the torque,as well as the dynamics that govern the response in the application of the forces and thetorque. In the bibliography, the delay in the application of the forces and the torque,was considered as a kind of disturbance and it was omitted from the control equations.A heuristic is also proposed for an accurate manipulation of the thrust. The modelpredictive controller has shown superior performance compared to the other controllers.Simulation results, in the presence of realistic environmental disturbances, are given thatdemonstrate the performance of each controller. Finally, a comparative study help usbetter understand how this system reacts under the governance of different controllers.The introduction of the performance indicator enriched our comparative study.The diploma thesis is organized in 7 chapters. In brief, chapter 1 is the introductorychapter which gives us the motivation of the work, chapters 2 and 3 describe the mechan-ical properties and the modelling of the platform. Chapter 4 focuses on the backsteppingcontroller and chapter 5 on the model predictive controller. Chapter 6 provides compar-ative remarks about the performance of the controllers and chapter 7 is the conclusivechapter. Our contributions are presented in chapters 4, 5 and 6.