Achieving robust and reliable autonomous operation even in complex unstructured environments is a central goal of field robotics. As the environments and scenarios to which robots are applied have continued to grow in complexity, so has the challenge of properly defining preferences and tradeoffs between various actions and the terrains they result in traversing. These definitions and parameters encode the desired behavior of the robot; therefore their correctness is of the utmost importance. Current manual approaches to creating and adjusting these preference models and cost functions have proven to be incredibly tedious and time-consuming, while typically not producing optimal results except in the simplest of circumstances. This thesis presents the development and application of machine learning techniques that automate the construction and tuning of preference models within complex mobile robotic systems. Utilizing the framework of inverse optimal control, expert examples of robot behavior can be used to construct models that generalize demonstrated preferences and reproduce similar behavior. Novel learning from demonstration approaches are developed that offer the possibility of significantly reducing the amount of human interaction necessary to tune a system, while also improving its final performance. Techniques to account for the inevitability of noisy and imperfect demonstration are presented, along with additional methods for improving the efficiency of expert demonstration and feedback. The effectiveness of these approaches is confirmed through application to several real world domains, such as the interpretation of static and dynamic perceptual data in unstructured environments and the learning of human driving styles and maneuver preferences. Extensive testing and experimentation both in simulation and in the field with multiple mobile robotic systems provides empirical confirmation of superior autonomous performance, with less expert interaction and no hand tuning. These experiments validate the potential applicability of the developed algorithms to a large variety of future mobile robotic systems.

[pt] Problemas de multifluxos surgem como modelos básicos no contexto de várias aplicações de fluxos em redes, tais como redes de telecomunicações, redes de transporte e logística. Em tais aplicações, os fluxos que atravessam a rede compartilham simultaneamente os mesmos recursos disponíveis e são definidos por suas próprias restrições. A cada uma das arestas ligando os pontos da rede está associado um custo, fixo ou proporcional à sua utilização. Este trabalho trata problemas de projeto de redes multifluxos, em que os custos estão associados às capacidades instaladas nas arestas. Particularmente, será estudado o caso em que a função de custo nas arestas possui o comportamento de uma função escada crescente e descontínua, para o qual métodos exatos de resolução são ineficientes. Métodos heurísticos são propostos para a resolução aproximada do problema e sintetizados em um algoritmo de multi-partida com memória adaptativa. Um mecanismo de intensificação, conhecido na literatura como construção de vocabulário, é também explorado e aplicado. Finalmente, experimentos computacionais são realizados e o método de resolução proposto é analisado quanto aos seus resultados e os resultados obtidos pelo método de resolução proposto são analisados. O método obtém as melhores soluções conhecidas para algumas instâncias da literatura. / [en] Multicommodity flow problems arise widely as basic models in the context of network flows applications such as telecommunication networks, transportation problems, and logistic. In these applicatons, the flows that cross the networks share the same avaiable resources simultaneously and are defined by their own constraints. Each edge connecting two nodes in the network has an associated cost that is either fixed or proportional to its use. This work focuses on a network design problem in which the cost are associated with the capacities installed in the edges. Particularly, the network design problem studied has discrete and step increasing cost functions on the edges, for which exact methods are inefficient. Heuristics are proposed for the approximate memory algorithm. An intensification mechanism, known in the literature as vocabulary building, is also explored and applied. Finally, computational experiments are performed and the results obtained with the proposed solution method are evaluated. The method obtains the best known solutions for some instances in the literature.

Questa tesi ha come obiettivo quello di contribuire all'analisi empirica della struttura dei costi del ciclo dei rifiuti solidi urbani in un campione di 534 comuni italiani, per un periodo di sette anni, dal 2004 al 2010. Dopo una panoramica sia delle varie fasi del servizio RSU che del funzionamento del settore in Italia, nel primo capitolo, il secondo analizza la letteratura empirica delle funzioni di costo, con particolare attenzione per i lavori più recenti, tra cui anche i pochissimi italiani sul tema. L’argomento del terzo capitolo riguarda lo studio della presenza di economie di scala e di densità nel settore dei rifiuti utilizzando una specificazione translogaritmica per la funzione di costo stimata con il metodo SUR. L'ultimo capitolo si propone di fornire risposte a domande molto attuali come: la raccolta porta a porta è più costosa? E' meglio smaltire in discarica i rifiuti residui o bruciarli? Data una certa quantità di rifiuti da smaltire, esiste una quantità ottimale da incenerire? Inoltre, dato che molte comunità usano contemporaneamente diversi siti di smaltimento, qual è il numero ottimale di siti per portare i propri rifiuti indifferenziati? / This thesis aims to contribute to the empirical analysis of the cost structure of municipal solid waste management in a sample 534 Italian municipalities, for a period of seven years, from 2004 to 2010. After giving an overview both of the various phases of the MSW service and of the functioning of the sector in Italy in the first chapter, the second chapter reviews the cost function studies, with particular emphasis on recent years including also the very few Italian papers on the topic. This topic of the third chapter the study of the presence of scale and density economies in the Italian waste sector using a translogarithmic specification estimated with the SUR method. The last chapter aims to provide answers to very actual questions as: is the door to door collection more expensive? Is it better to burry residual waste or to incinerate it? Given a certain amount of garbage, is there an optimal quantity to be incinerated? Moreover, given that many communities use contemporaneously different dumpsites, which is the perfect number of disposal sites for unsorted waste?

The aim of diploma thesis is to design technology components production orders. For a given part of the technology will be used non conventional deep drawing Wheelon method. For manufacturing lengthwise recesses and consequent cutting of shaped holes using a plasma cutting. The suitability of technological applications will be tested using the cost functions.

This thesis explores the possibility of using an optimal control scheme called Model Predictive Control (MPC), to control climatization systems for electric vehicles. Some components of electric vehicles, for example the batteries and power electronics, are sensitive to temperature and for this reason it is important that their temperature is well regulated. Furthermore, like all vehicles, the cab also needs to be heated and cooled. One of the weaknesses of electric vehicles is their range, for this reason it is important that the temperature control is energy efficient. Once the range of electric vehicles is increased the down sides compared to traditional combustion engine vehicles decrease, which could lead to an increase in the usage of electric vehicles. This could in turn lead to a decrease of greenhouse gas emission in the transportation sector. With the help of MPC it is possible for the controller to take more factors into consideration when controlling the system than just temperature and in this thesis the power consumption and noise are also taken into consideration. A simple model where parts of the climate system’s circuits were seen as point masses was developed, with nonlinear heat transfers occurring between them, which in turn were controlled by actuators such as fans, pumps and valves. The model was created using Simulink and MATLAB, and the MPC toolbox was used to develop nonlinear MPC controllers to control the climate system. A standard nonlinear MPC, a nonlinear MPC with custom cost functions and a PI controller where all developed and compared in simulations of a cooling scenario. The controllers were designed to control the temperatures of the battery, power electronics and the cab of an electric vehicle. The results of the thesis indicate that MPC could reduce power consumption for the climate control system, it was however not possible to draw any final conclusions as the PI controller that the MPC controllers were compared to was not well optimized for the system. The MPC controllers could benefit from further work, most importantly by applying a more sophisticated tuning method to the controller weights. What was certain was that it is possible to apply this type of centralized controller to very complex systems and achieve robustness without external logic. Even with the controller keeping track of six different temperatures and controlling 15 actuators, the control loop runs much faster than real time on a modern computer which shows promise with regard to implementing it on an embedded system.

Segmentação de objetos em imagens é um dos problemas mais fundamentais e desafiadores em processamento de imagem e visão computacional. O conhecimento de alto nível e específico do usuário é frequentemente requerido no processo de segmentação, devido à presença de fundos heterogêneos, objetos com bordas fracamente definidas, inomogeneidade de campo, ruído, artefatos, efeitos de volume parcial e seus efeitos conjuntos. Propriedades globais do objeto de interesse, tais como conexidade, restrições de forma e polaridade de borda, são conhecimentos prévios de alto nível úteis para a sua segmentação, permitindo a customização da segmentação para um objeto alvo. Nesse trabalho, apresentamos um novo método chamado Transformada Imagem-Floresta Orientada Conexa (COIFT, Connected Oriented Image Foresting Transform), que fornece soluções ótimas globais de acordo com uma medida de corte em grafo, incorporando a restrição de conexidade na Transformada Imagem-Floresta Orientada (OIFT, Oriented Image Foresting Transform), com o fim de garantir a geração de objetos conexos, bem como permitir o controle simultâneo da polaridade de borda. Enquanto o emprego de restrições de conexidade em outros arcabouços, tais como no algoritmo de corte-mínimo/fluxo-máximo (min-cut/max-flow), leva a um problema NP-difícil, a COIFT conserva o baixo custo computacional da OIFT. Experimentos mostram que a COIFT pode melhorar consideravelmente a segmentação de objetos com partes finas e alongadas, para o mesmo número de sementes em segmentação baseada em marcadores. / Object segmentation is one of the most fundamental and challenging problems in image processing and computer vision. The high-level and specific knowledge of the user is often required in the segmentation process, due to the presence of heterogeneous backgrounds, objects with poorly defined boundaries, field inhomogeneity, noise, artifacts, partial volume effects and their joint effects. Global properties of the object of interest, such as connectivity, shape constraints and boundary polarity, are useful high-level priors for its segmentation, allowing the customization of the segmentation for a given target object. In this work, we introduce a new method called Connected Oriented Image Foresting Transform (COIFT), which provides global optimal solutions according to a graph-cut measure in graphs, subject to the connectivity constraint in the Oriented Image Foresting Transform (OIFT), in order to ensure the generation of connected objects, as well as allowing the simultaneous control of the boundary polarity. While the use of connectivity constraints in other frameworks, such as in the min-cut/max-flow algorithm, leads to a NP-Hard problem, COIFT retains the low computational cost of OIFT. Experiments show that COIFT can considerably improve the segmentation of objects with thin and elongated parts, for the same number of seeds in segmentation based on markers.

Segmentação de objetos em imagens é um dos problemas mais fundamentais e desafiadores em processamento de imagem e visão computacional. O conhecimento de alto nível e específico do usuário é frequentemente requerido no processo de segmentação, devido à presença de fundos heterogêneos, objetos com bordas fracamente definidas, inomogeneidade de campo, ruído, artefatos, efeitos de volume parcial e seus efeitos conjuntos. Propriedades globais do objeto de interesse, tais como conexidade, restrições de forma e polaridade de borda, são conhecimentos prévios de alto nível úteis para a sua segmentação, permitindo a customização da segmentação para um objeto alvo. Nesse trabalho, apresentamos um novo método chamado Transformada Imagem-Floresta Orientada Conexa (COIFT, Connected Oriented Image Foresting Transform), que fornece soluções ótimas globais de acordo com uma medida de corte em grafo, incorporando a restrição de conexidade na Transformada Imagem-Floresta Orientada (OIFT, Oriented Image Foresting Transform), com o fim de garantir a geração de objetos conexos, bem como permitir o controle simultâneo da polaridade de borda. Enquanto o emprego de restrições de conexidade em outros arcabouços, tais como no algoritmo de corte-mínimo/fluxo-máximo (min-cut/max-flow), leva a um problema NP-difícil, a COIFT conserva o baixo custo computacional da OIFT. Experimentos mostram que a COIFT pode melhorar consideravelmente a segmentação de objetos com partes finas e alongadas, para o mesmo número de sementes em segmentação baseada em marcadores. / Object segmentation is one of the most fundamental and challenging problems in image processing and computer vision. The high-level and specific knowledge of the user is often required in the segmentation process, due to the presence of heterogeneous backgrounds, objects with poorly defined boundaries, field inhomogeneity, noise, artifacts, partial volume effects and their joint effects. Global properties of the object of interest, such as connectivity, shape constraints and boundary polarity, are useful high-level priors for its segmentation, allowing the customization of the segmentation for a given target object. In this work, we introduce a new method called Connected Oriented Image Foresting Transform (COIFT), which provides global optimal solutions according to a graph-cut measure in graphs, subject to the connectivity constraint in the Oriented Image Foresting Transform (OIFT), in order to ensure the generation of connected objects, as well as allowing the simultaneous control of the boundary polarity. While the use of connectivity constraints in other frameworks, such as in the min-cut/max-flow algorithm, leads to a NP-Hard problem, COIFT retains the low computational cost of OIFT. Experiments show that COIFT can considerably improve the segmentation of objects with thin and elongated parts, for the same number of seeds in segmentation based on markers.

Dans ce travail, la caractérisation mécanique de l’interphase entre les zones amorphes et cristallines dans le polyéthylène a été abordée. La caractérisation élastique est effectuée en appliquant deux approches micromécaniques à partir des données de la simulation moléculaire pour la zone interlamellaire. Ces approches micromécaniques sont d’une part le modèle étendu d’inclusion composite, et d’autre part la méthode de double inclusion. Les résultats des deux approches s’accordent parfaitement. Il a été mis en évidence que le tenseur de rigidité de l’interphase n’est pas défini positif, l’interphase est donc mécaniquement instable. La comparaison avec les résultats expérimentaux valide la méthodologie proposée. Pour la caractérisation hyperélastique, l’algorithme hybride proposé consiste à appliquer la loi de comportement d’un milieu continu isotrope, compressible et hyperélastique aux résultats de la simulation de la dynamique moléculaire d’un élément unitaire de polyéthylène. La notion d’optimisation d’un ensemble de fonctions coûts non négatives est l’idée clé de cette partie. Les paramètres hyperélastiques identifiés sont en bon accord avec ceux qui ont été estimés expérimentalement. L’évolution des frontières de l’interphase avec la déformation est le second résultat de cette analyse. La fin du travail est dédiée à la simulation numérique de la grande déformation viscoplastique d’un agrégat de polyéthylène. Le modèle de Gent adopté pour la contrainte de rappel, le tenseur de projection proposé pour l’approche modifiée de Taylor, et l’optimisation multiniveau font parties des contributions apportées. / Elastic characterization of the interphase layer in polyethylene is implemented by applying the relationships of two micromechanical approaches, “Extended Composite Inclusion Model” and “Double-Inclusion Method”, to the Monte Carlo molecular simulation data for the interlamellar domain. The results of the two approaches match perfectly. The interphase stiffness lacks the common feature of positive definiteness, which indicates its mechanical instability. Comparison with experimental results endorses the proposed methodology. For the hyperelastic characterization of the interlamellar domain and the interphase layer, the proposed hybrid algorithm consists in applying the constitutive equations of an isotropic, compressible, hyperelastic continuum to the molecular dynamics simulation results of a polyethylene stack. Evolution of the interphase boundaries are introduced as auxiliary variables and the notion of minimizing a set of nonnegative objective functions is employed for parameter identification. The identified hyperelastic parameters for the interlamellar domain arein good agreement with the ones that have been estimated experimentally. Finally, the large, viscoplastic deformation of an aggregate of polyethylene is reexamined. The Gent model adopted for the back stress of the noncrystalline phase, correcting the projection tensor for the modified Taylor approach, and the idea of multilevel optimization are among the contributions made.

The main objective of this thesis is to analyze the costs of a organization with fictitious name PROFIL Ltd. for the period 2007 - 2012 and then propose measures to improve the performance of the company by optimizing various cost types. The theoretical part describes the traditional tools and modern methods of cost management. The application part is focused on the characteristics of organization, evaluation of its financial situation during the reporting period, analysis of cost structures, analysis of break-even point, estimation of cost functions and analysis of organization?s individual cost items, followed by drawing conclusions and action plans.

A significant problem of functioning the public sector is to ensure the efficiency of public expenditures. Efficiency model's creating is a dynamic process that requires regular evaluation of the socioeconomic conditions. The efficiency of financial system and granting social services is still not measured in the Czech Republic. The area of social services is affected by number of criteria and factors. So it is necessary to apply different views on evaluating the efficiency of social care, social prevention and social counseling. The efficiency of social services can not be estimated as a whole, but we should focus on different groups of social services and individual services within the group. In this thesis the efficiency is measured in the two selected residental social care services -- homes for the elderly and homes for people with disabilities. The applied methods was used only to a limited extent because of lacks of many important variables in the area of official statics. In the case of multiple criteria methods the calculation was also impeded by the availability of suitable software.