Ph.D. Thesis by Nuno Alexandre Lopes Moreira da Cruz

Author: Nuno Alexandre Lopes Moreira da Cruz

Date: April 20, 2015

Ph.D. Committee:

Aurélio Campilho, Full Professor at FEUP (President)
Gerardo Acosta Lazo, Associate Professor, Universidad Nacional del Centro de la Provincia de Buenos Aires, Argentina
Yanwu Zhang, Senior Research Specialist, Monterey Bay Aquarium Research Institute, Moss Landing, USA
Fernando Lobo Pereira, Full Professor, Faculty of Engineering of the University of Porto
António Pedro Aguiar, Associate Professor, Faculty of Engineering of the University of Porto
Aníbal Matos, Assistant Professor, Faculty of Engineering of the University of Porto

Abstract:

Sampling the oceans is a very difficult task, not only requiring people to work in very uncomfortable and stressful conditions, but also because the equipment has to be reliable and robust. Very often, it also requires access to expensive support vessels. Moreover, traditional sampling methods can fail to capture the dynamics of many phenomena, due to insufficient sampling rate in space, in time, or both. There is therefore a great demand for faster, more effective and efficient methods for sampling the marine environment.

This thesis proposes an integrated approach to ocean sampling that includes the design and development of a new family of vehicles, the conception of innovative principles of operation and the implementation of the corresponding guidance algorithms.

Part of the proposal is a consistent program for the development of small size AUVs based on modular building blocks. This development relies on modularity both in terms of hardware construction, and also in terms of electronics, software and control. Using these blocks, the MARES AUV has been built, a man-portable, hovering AUV that has been continuously updated and used in the field in many different configurations, since 2007. In 2011, the versatility of the system components has been pushed further with the development of TriMARES, a 75kg, 3-body hybrid AUV/ROV system, which was developed in little over 6 months.

The family of new robotic platforms also includes small size ASVs based on modular components. Many of these building blocks are in fact common to the ones used for the assembly of AUVs, which emphasizes their value and versatility. They have been used to build the Zarco ASV, in 2005, and later to replicate the system to assemble Gama, in 2008. Both vehicles have been extensively used for testing multiple aspects of marine operations and they have undergone a continuous evolution since their first trials.

Finally, the work includes a strategy to use autonomous robotic platforms to efficiently map a class of oceanographic features. The distinctive characteristic of these features is that they have a boundary, a transition zone across which there is a much stronger variation in one of the values measured by the vehicle, as compared to the values in the rest of the survey area. Given the limited amount of computational power available on board, the strategy relies on admitting a relatively simple feature model for which the vehicle can estimate relevant parameters in real time using simple calculations, and then use these parameters to define the motion characteristics (and sampling pattern) accordingly. In order to implement this strategy, some algorithms are proposed for detecting the value and location of the maximum gradient of a scalar field traversed by the vehicle, as well as for the generation of motion references, in real time. This strategy was implemented in the vertical plane for one-dimensional feature tracking and also in the horizontal plane, for two-dimensional boundary tracking.