Abstract

Abstract
Previously, two municipal solid waste commercial composts (MSW1 and MSW2) were characterized. Although sharing the same type of raw material, most of their physicochemical, stability and maturity properties differed. The present study aimed to characterize them at a microbiological level, and to infer on possible relationships between the composts properties and the structure of their bacterial communities. Both the 16S rRNA gene-based PCR-DGGE profiling and 454-pyrosequencing technology showed that the structure of the bacterial communities of these composts was distinct. The bacterial community of MSW1 was more diverse than that of MSW2. Multivariate analyses revealed that the high electrical conductivity, Cu content as well as the low phytotoxity of compost MSW1, when compared to MSW2, contributed most to shape its bacterial community structure. Indeed, high abundance of halophilic (Halomonadaceae and Brevibacteriaceae) and metal resistant organisms (Brevibacteriaceae and Bacillaceae) were found in MSW1. In addition, Pseudonocardiaceae, Streptomycetaceae, Bacillaceae, and Brevibacteriaceae may have contributed to the high humic-like acids content and low phytotoxicity of MSW1. In contrast, the high organic matter content and the high density of the cultivable fungi population were the parameters most correlated with the structure of the bacterial community of compost MSW2, dominated by Corynebacteriaceae and mainly Aerococcaceae, taxonomic groups not commonly found in composts.

Abstract
Modelos virtuais 3D de edifícios são usualmente utilizados em áreas como a arquitetura e videojogos para fins de visualização de projetos de casas e povoamento de cenários virtuais, respetivamente. Tradicionalmente, a produção destes modelos requer mão-de-obra altamente especializada e consideráveis quantidades de tempo. Para abordar esta questão, muitos investigadores desenvolveram técnicas semiautomáticas para produzir modelos virtuais de forma expedita. Estas técnicas procedimentais providenciam diferentes formas de geração de edifícios, incluindo interiores e fachadas exteriores, que servem vários propósitos (por exemplo, geração de conteúdo para videojogos ou reconstruções arqueológicas). No entanto, as técnicas existentes com foco na construção de interiores normalmente só suportam a geração de plantas restritas por formas regulares ou polígonos de contorno obtidos a partir de conjuntos de retângulos. Ao mesmo tempo, a possibilidade de modelar quartos interiores através da especificação das suas paredes de restrição permanece pouco explorada. Além disso, a maioria das soluções de geração procedimental existentes recorrem a gramáticas complexas referentes aos aspetos geométricos, ou então, estruturas semânticas desenvolvidas para projetos com requisitos específicos, desconsiderando os standards desenvolvidos para ambientes urbanos virtuais, concretamente, CityGML. No sentido de abordar as questões indicadas, uma nova metodologia de modelação procedimental é proposta nesta tese, tendo como foco a produção de modelos virtuais de edifícios, incluindo exteriores circunscritos por formas arbitrárias e interiores formados por polígonos convexos. A regulação da metodologia é fornecida por uma ontologia para edifícios - que pode ser vista como um conjunto relacional de entidades baseadas em CityGML, extensíveis a estilos arquitetónicos específicos - através de várias estruturas de dados, tais como XML estruturado e gramática baseada na ontologia. Relativamente ao processo de suporte da repartição de espaço, uma abordagem treemap é usada para subdividir o layout representativo de uma dada base de edifício em subáreas inerentes a contentores e quartos interiores. Durante o desenvolvimento deste trabalho, diversas melhorias foram feitas ao treemap de forma progressiva, com o objetivo de permitir a subdivisão de diferentes tipos de polígonos de restrição que variam entre retângulos a formas arbitrárias. Além disso, na fase final deste trabalho, foi incorporado um método para a adaptação de paredes de quartos. Na sequência da subdivisão, vem um conjunto de operações que vai desde a marcação das transições até à extrusão das paredes que fornece o aspeto 3D. Também, uma abordagem estocástica experimental é proposta para automatizar a geração aleatória de edifícios, utilizando esta metodologia de modelação procedimental. Um conjunto de testes foi feito para demonstrar as capacidades da metodologia proposta na produção de edifícios com formatos distintos (edifícios limitados por formas convexas e não-convexas e quartos com um número específico de paredes de restrição) e diferentes estruturas arquitetónicas (casas de propósito geral, domus romanas) em curtos períodos de tempo. Além disso, a eficácia da abordagem treemap na subdivisão de layouts é mostrada, juntamente com um processo estocástico experimental para a geração automática de edifícios e também algumas medições de desempenho computacional.;
3D virtual models of buildings are commonly used in areas such as architecture and video games to preview a house project and to populate a virtual scenario, respectively. Traditionally, the production of these models requires highly skilled manpower and a considerable amount of time. To address this issue, many researchers have developed semi-automatic techniques to produce virtual models expeditiously. These procedural techniques provide different ways of generating buildings, including interiors and outer facades, to serve several purposes (e.g., content generation for video games or archaeological reconstruction). However, the existing techniques focusing on building interiors usually only support the generation of floor plans constrained by regular shapes or contour polygons obtained from rectangles sets. At the same time, the possibility of modelling interior rooms through the specification of its constraint walls remains poorly explored. Moreover, most of the existing procedural generation solutions are guided by complex grammars concerned with geometrical aspects or semantic structures that fit specific project requirements, apparently disregarding the established standards for virtual urban environments, specifically, CityGML. To overcome the noted issues, a novel procedural modelling methodology is proposed in this thesis, one that produces virtual models of buildings, including exteriors outlined by arbitrary shapes and interiors formed by convex polygons. Methodology’s regulation is provided by a building ontology - a CityGML-based knowledge structure, planned to be extensible to specific architecture styles - through several guiding data structures such as structured XML and ontology-based grammar. Regarding the supporting process, a treemap approach is used to subdivide the building layout into floor plan areas. During the development of this work, several improvements were progressively made to the treemap in order to enable the subdivision of different constraint polygon types which range from rectangles to arbitrary shapes. Moreover, in the most mature work stage, a method concerning inner room walls adaptation is addressed. Next, a set of operations is performed, from the marking transitions step to the extrusion process that provides the 3D aspect. In addition, an experimental stochastic approach is proposed to automate the production of random buildings using this procedural modelling methodology. A set of tests was made to demonstrate the capabilities of the proposed methodology in producing distinct building formats (buildings constrained by convex and nonconvex shapes, houses with specific room constraint walls) and different architectonic structures (general purpose houses, roman domus) in short time periods. Moreover, the effectiveness of the treemap approach in subdividing random layouts is shown, along with a generic stochastic process for automatic building generation and also some computational performance measurements.

Abstract

Abstract
Object representation is one of the most challenging tasks in robotics because it must provide reliable information in real-time to enable the robot to physically interact with the objects in its environment. To ensure robustness, a global object descriptor must be computed based on a unique and repeatable object reference frame. Moreover, the descriptor should contain enough information enabling to recognize the same or similar objects seen from different perspectives. This paper presents a new object descriptor named Global Orthographic Object Descriptor (GOOD) designed to be robust, descriptive and efficient to compute and use. We propose a novel sign disambiguation method, for computing a unique reference frame from the eigenvectors obtained through Principal Component Analysis of the point cloud of the target object view captured by a 3D sensor. Three principal orthographic projections and their distribution matrices are computed by exploiting the object reference frame. The descriptor is finally obtained by concatenating the distribution matrices in a sequence determined by entropy and variance features of the projections. Experimental results show that the overall classification performance obtained with GOOD is comparable to the best performances obtained with the state-of-the-art descriptors. Concerning memory and computation time, GOOD clearly outperforms the other descriptors. Therefore, GOOD is especially suited for real-time applications. The estimated object's pose is precise enough for real-time object manipulation tasks.

Abstract
In order to support the intensive development of agricultural crops and, in particular the floricultural inside a greenhouse, with the perspective of a quick distribution in the market, increasing the economic benefits and supported on efficient and intelligent management systems energy, it is mandatory to conceive a model based on Cyber-Physical Systems (CPS) This implies, accordingly, increases in renewable primary energy sources utilization coupled with sensing technologies, include developments on Internet of Things (IoT) and Cloud Computing (CC), supported with Information and Communication Technologies (ICT) that will lead to new architectural approach applied to a proposed energy system, based on a sustainable and more engineering autonomous process. This work comes up with a new energy model that retrofits the system of a greenhouse supported with multiple sensors in one grid, to expand into CPS concept to manage sensors and controllers that will improve a profitable energy system.