Un robot bipède doit pouvoir marcher en toute sécurité dans une foule. Pour cela, il faut prendre en compte deux aspects : l’équilibre et l'évitement des collisions. Maintenir l’équilibre implique d'éviter les défaillances dynamiques et cinématiques de la dynamique instable du robot. Pour ce qui est de l’évitement des collisions, il s’agit d’éviter le contact entre le robot et des individus. Nous voulons être capables de satisfaire ces deux contraintes simultanément, à l’instant présent mais aussi dans le futur. Nous pouvons assurer l’équilibre du robot indéfiniment en le faisant entrer dans un cycle limite de marche ou en le faisant s’arrêter après quelques pas. Néanmoins, une telle garantie pour l’évitement d’obstacle n’est pas possible pour plusieurs raisons : impossibilité de connaître de manière absolue la direction vers laquelle les individus se dirigent, limitations cinématiques et dynamiques du robot, mouvement adverse de la foule, etc. Nous traitons ces limitations avec une stratégie standard de navigation dans une foule, appelée passive safety, qui nous permet de formuler une loi de commande prédictive avec laquelle nous assurons l’équilibre et l'évitement des collisions, de manière unifiée, en faisant s’arrêter le robot de manière sécurisée et en temps fini. De plus, nous définissons une nouvelle stratégie de navigation sûre basée sur le principe d’évitement des collisions aussi longtemps que possible, qui a la propriété de minimiser leur apparition et sévérité. Nous proposons une formulation lexicographique qui synthétise des mouvements conformes à ce principe. Nous augmentons les degrés de liberté de la locomotion d’un robot bipède en permettant la variation de l’orientation et de la durée des pas en ligne. Cependant, cela introduit des non-linéarités dans les contraintes de nos problèmes d’optimisation. Nous faisons des approximations de ces contraintes non-linéaires avec des contraintes linéaires sûres de sorte que la satisfaction des secondes implique la satisfaction des premières. Nous proposons une nouvelle méthode de résolution des problèmes non-linéaires (Optimisation Quadratique Successive Sûre) qui assure la faisabilité des itérations de Newton en utilisant cette redéfinition des contraintes. Nous simulons la marche d’un robot bipède dans une foule pour évaluer la performance de nos lois des commandes. D’une part, nous réussissons à réduire (statistiquement) la quantité et la sévérité des collisions en comparaison avec la méthode de passive safety, spécialement dans les conditions d’incertitude de la marche du robot dans une foule. D’autre part, nous montrons des exemples de comportements typiques du robot, qui découlent de la liberté de choisir l’orientation et la durée des pas. Nous rapportons le coût de calcul de notre méthode de résolution des problèmes non-linéaires en comparaison avec une méthode standard. Nous montrons qu’une seule itération de Newton est nécessaire pour arriver à une solution faisable, mais que le coût de calcul dépend du nombre de factorisations de l’active set dont nous avons besoin pour arriver à l’active set optimal. / We want a biped robot to walk safely in a crowd. This involves two aspects: balance and collision avoidance. The first implies avoiding kinematic and dynamical failures of the unstable walking dynamics of the robot; the second refers to avoiding collisions with people. We want to be able to solve both problems not only now but also in the future. We can ensure balance indefinitely by entering in a cyclic walk or by making the robot stop after a couple of steps. Nonetheless, we cannot give a comparable guarantee in collision avoidance for many reasons: impossibility of having absolute knowledge of where people are moving, kinematic/dynamical limitations of the robot, adversarial crowd motion, etc. We address this limitation with a standard strategy for crowd navigation, known as passive safety, that allows us to formulate a unified Model Predictive Control approach for balance and collision avoidance in which we require the robot to stop safely in finite time. In addition, we define a novel safe navigation strategy based on the premise of avoiding collisions for as long as possible that minimizes their occurrence and severity. We propose a lexicographic formulation that produces motions that comply with such premise.We increase the degrees of freedom of the locomotion of a biped robot by allowing the duration and orientation of its steps to vary online. This introduces nonlinearities in the constraints of the optimization problems we solve. We approximate these nonlinear constraints with safe linear constraints so that satisfying the latter implies satisfying the former. We propose a novel method (Safe Sequential Quadratic Programming) that ensures feasible Newton iterates in the solution of nonlinear problems based on this redefinition of constraints.We make a series of simulations of a biped robot walking in a crowd to evaluate the performance of our proposed controllers. We are able to attest the reduction in the number and in the severity of collisions with our proposed navigation strategy in comparison with passive safety, specially when there is uncertainty in the motion of people. We show typical behaviors of the robot that arise when we allow the online variation of the duration and orientation of the steps and how it further improves collision avoidance. We report the computational cost of our proposed numerical method for nonlinear problems in comparison with a standard method. We show that we only need one Newton iteration to arrive to a feasible solution but that the CPU time is dependent on the amount of active set factorizations needed to arrive to the optimal active set.

In this work we study the stochastic optimal control problem of discrete-time linear systems subject to Markov jumps and multiplicative noises. We consider the multiperiod and finite time horizon optimization of a mean-variance cost function under a new criterion. In this new problem, we apply a constraint on the total output variance weighted by its risk parameter while maximizing the expected output. The optimal control law is obtained from a set of interconnected Riccati difference equations, extending previous results in the literature. The application of our results is exemplified by numerical simulations of a portfolio of stocks and a risk-free asset. / Neste trabalho, estudamos o problema do controle ótimo estocástico de sistemas lineares em tempo discreto sujeitos a saltos Markovianos e ruídos multiplicativos. Consideramos a otimização multiperíodo, com horizonte de tempo finito, de um funcional da média-variância sob um novo critério. Neste novo problema, maximizamos o valor esperado da saída do sistema ao mesmo tempo em que limitamos a sua variância total ponderada pelo seu parâmetro de risco. A lei de controle ótima é obtida através de um conjunto de equações de diferenças de Riccati interconectadas, estendendo resultados anteriores da literatura. São apresentadas simulações numéricas para uma carteira de investimentos com ações e um ativo de risco para exemplificarmos a aplicação de nossos resultados.

Neste trabalho foi estudado um problema de controle de sistemas lineares com saltos Markovianos sem observação da variável de salto, que pode ser escrito como um problema de otimização de considerável complexidade. As contribuições para a área estão divididas em três aspectos. Um dos avanços foi a elaboração de um contraexemplo para a conjectura de que há somente um mínimo local isolado para o problema. Além disso, foi estudado o problema de otimização intermediário, que consiste em fixar todas as variáveis do problema exceto duas matrizes de ganhos, e os resultados indicam que, com uma pequena alteração na formulação, este é um problema biquadrático. Por fim, novos algoritmos foram elaborados a partir de um método disponível na literatura, chamado de método Variacional, adaptando-o para atualizar os ganhos aos pares, levando a problemas intermediários biquadráticos. Três métodos foram implementados para a resolução destes problemas: dois métodos clássicos de descida, Newton e Gradiente, e uma adaptação do próprio método Variacional. Para a análise dos resultados foram utilizados exemplos gerados aleatoriamente a partir do Gerador de SLSM, que pode ser encontrado na literatura, e o método Variacional como referência para comparação com os métodos propostos / This work addresses a control problem arising in linear systems with Markov jumps without observation of the jump variable and advances in three different aspects. First, it is presented a counterexample to the conjecture that states about the uniqueness of local minimum. Second, the intermediary optimization problem, which sets all the variables of the problem except two arrays of gains, was studied and the results suggested that a slight modification in the formulation makes the intermediary problem a biquadratic one. Finally, new algorithms were developed based on a method available in the literature, which is frequently referred to as the Variational method, adapting it to update the gains in pairs, leading to biquadratic intermediary problems. Three methods were implemented to solve these intermediary problems: two classical descent methods, Newton and Gradient, and an adaptation of the Variational method. To evaluate the performance of the proposed methods, randomly generated examples were used and the Variational method was set as reference for comparing the results

Nesta tese, apresentamos uma metodologia concreta para calcular os controles -ótimos para sistemas estocásticos não-Markovianos. A análise trajetória a trajetória e o uso da estrutura de discretização proposta por Leão e Ohashi [36] conjuntamente com argumentos de seleção mensuráveis, nos forneceu uma estrutura para transformar um problema infinito dimensional para um finito dimensional. Desta forma, garantimos uma descrição concreta para uma classe bastante geral de problemas. / In this thesis, we present a concrete methodology to calculate the -optimal controls for non-Markovian stochastic systems. A pathwise analysis and the use of the discretization structure proposed by Leão and Ohashi [36] jointly with measurable selection arguments, allows us a structure to transform an infinite dimensional problem into a finite dimensional. In this way, we guarantee a concrete description for a rather general class of stochastic problems.

"A number of outstanding problems in robotic motion and manipulation involve tasks where degrees of freedom (DoF), be they part of the robot, an object being manipulated, or the surrounding environment, cannot be accurately controlled by the actuators of the robot alone. Rather, they are also controlled by physical properties or interactions - contact, robot dynamics, actuator behavior - that are influenced by the actuators of the robot. In particular, we focus on two important areas of poorly controlled robotic manipulation: motion planning for deformable objects and in deformable environments; and manipulation with uncertainty. Many everyday tasks we wish robots to perform, such as cooking and cleaning, require the robot to manipulate deformable objects. The limitations of real robotic actuators and sensors result in uncertainty that we must address to reliably perform fine manipulation. Notably, both areas share a common principle: contact, which is usually prohibited in motion planners, is not only sometimes unavoidable, but often necessary to accurately complete the task at hand. We make four contributions that enable robot manipulation in these poorly controlled tasks: First, an efficient discretized representation of elastic deformable objects and cost function that assess a ``cost of deformation' for a specific configuration of a deformable object that enables deformable object manipulation tasks to be performed without physical simulation. Second, a method using active learning and inverse-optimal control to build these discretized representations from expert demonstrations. Third, a motion planner and policy-based execution approach to manipulation with uncertainty which incorporates contact with the environment and compliance of the robot to generate motion policies which are then adapted during execution to reflect actual robot behavior. Fourth, work towards the development of an efficient path quality metric for paths executed with actuation uncertainty that can be used inside a motion planner or trajectory optimizer."

In this thesis, we present some contributions to the Markov jump linear systems theory in a context of partial information on the Markov chain. We consider that the state of the Markov chain cannot be measured, but instead there is only an observed variable that could model an asynchronous phenomenon between the application and the plant, or a simple fault detection and isolation device. In this formulation, we investigate the problem of designing controllers and filters depending only on the observed variable in the context of H2, H?, and mixed H2/H? control theory. Numerical examples and academic applications are presented for active-fault tolerant control systems and networked control systems. / Nesta tese, apresentamos algumas contribuições para a teoria de sistemas lineares com saltos markovianos em um contexto de observação parcial da cadeia de Markov. Consideramos que o estado da cadeia de Markov não pode ser medido, porém existe uma variável observada que pode modelar um fenômeno assíncrono entre a aplicação e a planta, ou ainda um dispositivo de detecção de falhas simples. Através desse modelo, investigamos o problema da síntese de controladores e filtros que dependem somente da variável observada no contexto das teorias de controle H2, H?, e misto H2/H?. Exemplos numéricos e aplicações acadêmicas são apresentadas no âmbito dos sistemas de controle tolerantes a falhas e dos sistemas de controle através da rede.

Esta dissertação visa a modelagem, simulação e controle de um veículo aéreo não tripulado (VANT) do tipo quadricóptero, utilizando-se as técnicas de controle ótimo e controle robusto no espaço de estados. O quadricóptero deve realizar as funções de decolagem, voo em cruzeiro e pouso de maneira autônoma. A dissertação inclui a síntese e análise comparativa entre um observador de estados de ordem plena de Luenberger e um filtro de Kalman. Além disso, um controlador linear quadrático gaussiano e um controlador robusto serão sintetizados e avaliados, procurando-se avaliar qual tem o melhor desempenho nas diversas tarefas do VANT. / This dissertation includes the modeling, simulation and control of a quadrotor unmanned aerial vehicle by using optimum control and robust control techniques on the space state. The quadrotor must perform the takeoff, cruise flight and landing in an autonomous way. This report also presents the synthesis and comparative analysis between a Luenberger full order state observer and a Kalman filter. A linear quadratic gaussian controller and a robust controller will be also synthetized and analyzed, to compare which one exhibits the best performance on the UAV tasks.

La qualité des prévisions météorologiques repose principalement sur la qualité du modèle utilisé et de son état initial. Cet état initial est reconstitué en combinant les informations provenant du modèle et des observations disponibles en utilisant des techniques d'assimilation de données. Historiquement, les prévisions et l'assimilation sont réalisées dans l'atmosphère et l'océan de manière découplée. Cependant, les centres opérationnels développent et utilisent de plus en plus des modèles couplés océan-atmosphère. Or, assimiler des données de manière découplée n'est pas satisfaisant pour des systèmes couplés. En effet, l'état initial ainsi obtenu présente des inconsistances de flux à l'interface entre les milieux, engendrant des erreurs de prévision. Il y a donc besoin d'adapter les méthodes d'assimilation aux systèmes couplés. Ces travaux de thèse s'inscrivent dans ce contexte et ont été effectués dans le cadre du projet FP7 ERA-Clim2, visant à produire une réanalyse globale du système terrestre.Dans une première partie, nous introduisons les notions d'assimilation de données, de couplage et les différentes méthodologies existantes appliquées au problème de l'assimilation couplée. Ces méthodologies n’étant pas satisfaisantes en terme de qualité de couplage ou de coût de calcul, nous proposons, dans une seconde partie, des méthodes alternatives. Nous faisons le choix de méthodes d'assimilation basées sur la théorie du contrôle optimal. Ces alternatives se distinguent alors par le choix de la fonction coût à minimiser, des variables contrôlées et de l’algorithme de couplage utilisé. Une étude théorique de ces algorithmes a permis de déterminer un critère nécessaire et suffisant de convergence dans un cadre linéaire. Pour conclure cette seconde partie, les performances des différentes méthodes introduites sont évaluées en terme de qualité de l’analyse produite et de coût de calcul à l’aide d’un modèle couplé linéaire 1D. Dans une troisième et dernière partie, un modèle couplé non-linéaire 1D incluant des paramétrisations physique a été développé et implémenté dans OOPS (textit{Object-Oriented Prediction System}) qui est une surcouche logicielle permettant la mise en œuvre d’un ensemble d’algorithmes d’assimilation de données. Nous avons alors pu évaluer la robustesse de nos algorithmes dans un cadre plus réaliste, et conclure sur leurs performances vis à vis de méthodes existantes. Le fait d’avoir développé nos méthodes dans le cadre de OOPS devrait permettre à l’avenir de les appliquer aisément à des modèles réalistes de prévision. Nous exposons enfin quelques perspectives d'amélioration de ces algorithmes. / In the context of operational meteorology and oceanography, forecast skills heavily rely on the model used and its initial state. This initial state is produced by a proper combination of model dynamics and available observations via data assimilation techniques. Historically, numerical weather prediction is made separately for the ocean and the atmosphere in an uncoupled way. However, in recent years, fully coupled ocean-atmosphere models are increasingly used in operational centres. Yet the use of separated data assimilation schemes in each medium is not satisfactory for coupled problems. Indeed, the result of such assimilation process is generally inconsistent across the interface, thus leading to unacceptable artefacts. Hence, there is a strong need for adapting existing data assimilation techniques to the coupled framework. This PhD thesis is related to this context and is part of the FP7 ERA-Clim2 project, which aim to produce an earth system global reanalysis.We first introduce data assimilation and model coupling concepts, followed by some existing algorithms of coupled data assimilation. Since these methods are not satisfactory in terms of coupling strengh or numerical cost, we suggest, in a second part, some alternatives. These are based on optimal control theory and differ by the choice of the cost function to minimize, controled variable and coupling algorithm used. A theoretical study of these algorithms exhibits a necessary and sufficient convergence criterion in a linear case. To conclude about this second part, the different methods are compared in terms of analysis quality and numerical cost using a 1D linear model. In a third part, a 1D non-linear model with subgrid parametrizations was developed and implemented in OOPS (Object-Oriented Prediction System), a software overlay allowing the implementation of a set of data assimilation algorithms. We then assess the robustness of the different algorithms in a more realistic case, and concluded about their performances against existing methods. By implementing our methods in OOPS, we hope it should be easier to use them with operational forecast models. Finally, we expose some propects for improving these algorithms.

Inicialmente, os conceitos fundamentais e a problemática básica subjacentes ao Controle de Sistemas Quânticos são apresentados, destacando-se, por exemplo, as questões físicas e dinâmicas envolvidas, os principais tipos e metodologias de controle no contexto quântico, bem como aplicações existentes e potenciais de Controle Quântico, muitas das quais situando-se na vanguarda da Ciência e da Tecnologia. Segue-se uma exposição do arcabouço teórico básico e do formalismo padrão da Mecânica Quântica, tendo em vista prover os elementos necessários à compreensão de sistemas quânticos, sua dinâmica e seu controle. O conceito de Controlabilidade é, então, apresentado no contexto de Sistemas Quânticos. Em seqüência, os fundamentos do Controle Quântico Ótimo são desenvolvidos como uma extensão da Teoria Clássica de Controle Ótimo, apresentando-se exemplos de aplicações. Ao problema da transferência de estados quânticos para um estado-alvo em tempo mínimo é devotada especial atenção, dada sua grande relevância em aplicações tecnológicas de ponta, como em Computação Quântica e Processamento de Informação Quântica. A partir de limitações físicas que são inerentes a qualquer sistema quântico, no tocante ao tempo mínimo necessário para que ocorra uma transição de estados, propõem-se Fatores de Mérito para quantificar a eficiência dos controles quânticos ótimos que minimizam o tempo de transferência de estados. Exemplos de aplicação, estudos teóricos e estudos de casos são levados a cabo para a definição dos Fatores de Mérito associados. Este trabalho termina com estudos relativos a uma possível formulação da Teoria de Controle Quântico Ótimo em termos de Integrais de Trajetória para o tratamento de sistemas quânticos contínuos, em especial, o controle espaço-temporal de partículas quânticas. Um possível emprego do Efeito Aharonov-Bohm é também discutido como estratégia de Controle Quântico. / Firstly, the fundamental concepts and the basic issues concerning the Control of Quantum Systems are presented, highlighting, for example, related physical and dynamical questions, the main control types and methodologies in the quantum context, as well as current and potential applications of Quantum Control, many of them situated on the avant-garde of Science and Technology. Then follows an exposition of the basic theoretical framework and the standard formalism of Quantum Mechanics, whose aim is to provide the necessary elements for understanding quantum systems, quantum dynamics and control. The concept of Controlability is then presented in the context of Quantum Systems. Subsequently, the fundamental concepts of Quantum Optimal Control are developed as an extension of the Classical Optimal Control Theory, featuring some examples of application. To the problem of transfering quantum states to a certain target state at minimal time a special attention is devoted, having in mind its great relevance in state-of-art technological applications, e.g., Quantum Computation and Quantum Information Processing. From physical limitations that are inherent to any quantum systems, regarding the minimal time necessary to perform a state transition, one proposes Figures of Merit in order to quantify the efficiency of optimal quantum controls which minimize the state transfer time. Examples of applications, theoretical studies and case studies are carried out in order to define the associated Figures of Merit. This work ends with studies concerning a possible formulation of Optimal Quantum Control Theory in terms of Path Integrals for handling continuous quantum systems, particularly, the space-time control of quantum particles. A possible use of the Aharonov-Bohm Effect is also discussed as a Quantum Control strategy.

Este trabalho trata do planejamento de trajetórias que minimizam as perdas elétricas no KA\'I yxo, um robô escalador de árvores que tem por finalidade realizar monitoramento ambiental em florestas através da coleta de diferentes tipos de dados. Como essa aplicação requer que o robô permaneça em ambientes remotos, o estudo de técnicas que reduzam as perdas de energia a fim de que se aumente o tempo em operação do robô se mostra relevante, sendo a minimização das perdas elétricas uma contribuição importante nesse sentido. Estruturalmente, o KA\'I yxo consiste em um robô bípede com duas garras e quatro ligamentos interconectados por três juntas rotacionais. Além disso, seu mecanismo de andadura foi biologicamente inspirado na forma de locomoção observada em lagartas mede-palmos, o que permitiu tratar o robô como um manipulador industrial, cuja base é o ligamento associado à garra engastada e cujo efetuador é o ligamento associado à garra livre. Com isso, quando conveniente, o robô foi tratado em dois casos, conforme a garra que se encontra engastada. Inicialmente, realizou-se a modelagem matemática do robô, obtendo-se as equações cinemáticas direta e inversa, e dinâmicas, bem como o modelo das juntas segundo a abordagem do controle independente por junta. Posteriormente, formulou-se um problema de controle ótimo, solucionado através de um método numérico que o transformou em um problema de programação quadrática, que por sua vez foi resolvido iterativamente. Por fim, as trajetórias ótimas planejadas foram implementadas no robô real e, como forma de validação, as novas perdas elétricas foram comparadas com as das trajetórias anteriormente executadas pelo robô, determinando-se a correspondente economia de energia. / This work deals with the minimum-energy trajectory planning, related to the electrical losses, in KA\'I yxo, a tree-climbing robot that aims to perform environmental monitoring in forests through the collection of different types of data. As this application requires that the robot remains in remote environments, the study of techniques that reduce energy losses in order to increase the operation time of the robot is shown to be relevant, and the minimization of the electrical losses is an important contribution in this sense. Structurally, KA\'I yxo consists of a biped robot with two claws and four links interconnected by three revolute joints. In addition, its gait mechanism was biologically inspired in the form of locomotion observed in caterpillars, allowing to treat the robot as an industrial manipulator, which base is the link associated with the fixed claw and which end-effector is the link associated with the free claw. In consequence, when convenient, the robot was treated in two cases, according to the claw that is fixed. Initially, the mathematical model of the robot was developed, being obtained the forward and inverse kinematic and dynamic equations, as well as the model of the joints according to the independent joint control approach. Subsequently, an optimal control problem was formulated, which was solved through a numerical method that turned it into a quadratic programming problem, which in turn was solved iteratively. Finally, the planned optimal trajectories were implemented in the real robot and, as a form of validation, the new electrical losses were compared with those of the trajectories previously executed by the robot, being determined the corresponding energy saving.