Ce mémoire est consacré à la mise en œuvre de commandes d'un train de véhicules intelligents sur autoroute ayant pour objectifs principaux de réduire la congestion et d’améliorer la sécurité routière. Après avoir présenté l'état de l'art sur des systèmes de conduite automatisée, des modèles de la dynamique longitudinale et latérale du véhicule sont présentés. Ensuite, des stratégies de contrôle longitudinal et latéral sont étudiées.D'abord, le contrôle longitudinal est conçu pour être hiérarchique avec un contrôleur de niveau supérieur et un contrôleur de niveau inférieur. Pour celui de niveau supérieur, une régulation d'inter-distance SSP (Safety Spacing Policy) est proposée. Nous avons constaté que la SSP peut assurer la stabilité de la chaîne et la stabilité des flux de trafic et augmenter ainsi la capacité de trafic. Puis, pour celui de niveau inférieur, une loi de commande floue coordonnée est proposée pour gérer l'accélérateur et le freinage. Ensuite, une loi de commande multi-modèle floue est conçue pour le contrôle latéral. De plus, pour réaliser des transformations lisses entre les différentes opérations latérales, une architecture de contrôle hiérarchique est proposée. Puis, l'intégration des commandes longitudinale et latérale est étudiée. Enfin, l'estimation des variables d’états du véhicule est discutée. Un filtre de Kalman-Bucy est conçu pour estimer les états du véhicule. En outre, un prototype de véhicule intelligent à échelle réduite est également présenté. Les performances des divers algorithmes de commande proposés ont été testées par simulations, et les résultats ont été confirmés par les premières expériences en utilisant le prototype / This PhD thesis is dedicated to the control strategies for intelligent vehicle platoon in highway with the main aims of alleviating traffic congestion and improving traffic safety. After a review of the different existing automated driving systems, the vehicle longitudinal and lateral dynamic models are derived. Then, the longitudinal control and lateral control strategies are studied respectively. At first, the longitudinal control system is designed to be hierarchical with an upper level controller and a lower level controller. For the upper level controller, a safety spacing policy (SSP) is proposed. It is shown that the proposed SSP can ensure string stability, traffic flow stability and improve traffic capacity. Then, a coordinated throttle and brake fuzzy controller (lower level controller) is designed, in which a logic switch is designed to coordinate the two actuators (throttle and brake pedals). Second, for the lateral control, a multi-model fuzzy controller is designed. And a hierarchical lateral control architecture is also proposed, which can effectuate flexible switch between different lateral operations. After that, the integration of the longitudinal controller and lateral controller is also studied. Finally, the estimation of vehicle states is discussed. A Kalman-Bucy filter is designed to estimate vehicle states in lateral dynamics. Moreover, a reduced scale multi-sensor intelligent vehicle prototype is also presented. The performances of the divers control algorithms proposed in this thesis have been tested in numerical simulations, and the first step experiments with the reduced scale vehicle prototype gave encouraging results

Mit dem Übergang von bauwerksorientierter zu prozessorientierter Arbeitsweise erlangt die domänenübergreifende Bereitstellung von Informationen wachsende Bedeutung. Das betrifft bspw. die Erstellung von Controlling-Kennwerten, die Vorbereitung von Simulationen oder die Betrachtung neuer Aspekte wie Energieeffizienz. Aktuelle Datenformate und Erschließungsmethoden können diese Herausforderung jedoch nicht befriedigend bewältigen. Daher bedarf es einer Methode, welche interdisziplinäre Bauinformationsprozesse uneingeschränkt ermöglicht. Vorhandene Kommunikationsprozesse und Fachanwendungen sollen dabei beibehalten und weitergenutzt werden können. Mit der Multimodell-Methode wird ein Lösungsansatz für die strukturellen Probleme interdisziplinärer Bauinformationsprozesse vorgestellt. Multimodelle bündeln heterogene Fachmodelle unterschiedlicher Domänen und erlauben die Verbindung ihrer Elemente in externen, ID-basierten Linkmodellen. Da die Fachmodelle unberührt bleiben, wird auf diesem Weg eine lose und temporäre Kopplung ermöglicht. Durch den Verzicht auf ein führendes oder integrierendes Datenschema werden keine Transformationsprozesse benötigt, können etablierte und heute übliche Datenformate weitergenutzt und die verlinkten Fachmodelle neutral ausgetauscht werden. Die in Multimodellen verknüpften Daten bieten einen informationellen Mehrwert gegenüber alleinstehenden Fachmodellen. Zusammengehörende Informationen können über die persistenten Links automatisch ausgewertet werden, anstelle manuell vom Menschen immer wieder flüchtig neu zugeordnet werden zu müssen. Somit erscheint ein Multimodell gegenüber einem Benutzer wie ein einziger abgeschlossener Informationsraum. Um solche datenmodell-, datenformat- und domänenübergreifenden Informationsräume komfortabel erstellen und filtern zu können, wird die deklarative Multimodell-Abfragesprache MMQL eingeführt. Diese erlaubt einen generischen Zugriff auf die Originaldaten und bildet die Kernkonzepte der Multimodell-Erschließung - mehrwertige Linkerzeugung und strukturelle Linksemantik - ab. Ein zugehöriger Interpreter ermittelt den Lösungsweg für konkrete Anweisungen und führt diesen auf realen Daten aus. Die Umsetzung und Bereitstellung der Konzepte als IT-Komponenten auf verschiedenen Ebenen - von der Datenstruktur über Bibliotheken und Services bis hin zur alleinstehenden, universellen Multimodell-Software M2A2 - erlaubt die sofortige und direkte Anwendung der Multimodell-Methode in der Praxis. / With the transition of building-oriented to process-oriented work, the provision of cross-domain information gained growing importance - for example in the creation of controlling parameters, the preparation of simulations or when considering new aspects such as energy efficiency. However, current data formats and access methods cannot cope with this challenge satisfactory. Therefore, a method is required, that enables interdisciplinary construction information processes fully. Thereby existing communication processes and domain applications have to be retained and continued to be used as possible. With the multi-model method, an approach to structural problems of such interdisciplinary construction information processes is presented. Multi-models combine heterogeneous models of different domains and allow the connection of their elements in external ID-based link models. As the domain models remain unaffected, a loose and temporary coupling is possible in this way. By not using a leading or integrating data schema, no transformation processes are required, common established data formats can be retained and the linked domain models can be exchanged neutrally. The linked data in multi-models offer an additional value of information over single domain models. Information belonging together can be automatically evaluated by the persistent links - instead of being repeatedly reassigned by people in a volatile way. Thus, a multi-model appears to a user as a single self-contained information space. In order to create and filter such cross-format and cross-domain information spaces comfortably, the declarative multi-model query language MMQL is introduced. It allows for generic access to the original data and integrates the core concepts of the multi-model development - n-ary link generation and structural link semantics. An associated interpreter determines the approach for specific instructions and executes it on real data. The implementation and deployment of the concepts as IT components at various levels - from the data structure via libraries and services, to the universal multi-model software M2A2 - allows an immediate and direct application of the multi-model method in practice.

Near Infrared (NIR) and Photoacoustic (PA) Imaging are promising imaging modalities that provides functional information of the soft biological tissues in-vivo, with applica-tions in breast and brain tissue imaging. These techniques use near infrared light in the wavelength range of (600 nm - 900 nm), giving an advantage of being non-ionizing imaging modality. This makes the prolong bed-side monitoring of tissue feasible, making them highly desirable medical imaging modalities in the clinic. The computation models that are deployed in these imaging scenarios are computationally demanding and often require a high performance computing systems to deploy them in real-time. This the-sis presents three computationally e cient models for near-infrared and photoacoustic imaging, without compromising the quality of measured functional properties, to make them more appealing in clinical scenarios. The attenuation of near-infrared (NIR) light intensity as it propagates in a turbid medium like biological tissue is described by modi ed the BeerLambert law (MBLL). The MBLL is generally used to quantify the changes in tissue chromophore concen-trations for NIR spectroscopic data analysis. Even though MBLL is e ective in terms of providing qualitative comparison, it su ers from its applicability across tissue types and tissue dimensions. A Lambert-W function-based modeling for light propagation in biological tissues is proposed and introduced, which is a generalized version of the Beer-Lambert model. The proposed modeling provides parametrization of tissue properties, which includes two attenuation coe cients o and . The model is validated against the Monte Carlo simulation, which is the gold standard for modeling NIR light propagation in biological tissue. Numerous human and animal tissues are included to validate the proposed empirical model, including an inhomogeneous adult human head model. The proposed model, which has a closed form (analytical), is rst of its kind in providing accurate modeling of NIR light propagation in biological tissues. Model based image reconstruction techniques yield better quantitative accuracy in photoacoustic (PA) image reconstruction, especially in limited data cases. An exponen-tial ltering of singular values is proposed for carrying out the image reconstruction in photoacoustic tomography. The results were compared with widely popular Tikhonov regularization, time reversal, and the state of the art least-squares QR based reconstruc-tion algorithms for three digital phantom cases with varying signal-to-noise ratios of data. The exponential ltering provided superior photoacoustic images of better quanti-tative accuracy. Moreover, the proposed ltering approach was observed to be less biased towards regularization parameter and did not come with any additional computational burden as it was implemented within the Tikhonov ltering framework. It was also shown that the standard Tikhonov ltering becomes an approximation to the proposed exponential ltering. The model based image reconstruction techniques for photoacoustic tomography re-quire an explicit regularization. An error estimate minimization based approach was proposed and developed for the determination of regularization parameter for PA imag-ing. The regularization was used within Lanczos bidiagonalization framework, which provides the advantage of dimensionality reduction for a large system of equations. The proposed method was computationally faster than the state of the art techniques and provided similar performance in terms of quantitative accuracy in reconstructed im-ages.The estimate can also be utilized in determining suitable regularization parameter for other popular techniques such as Tikhonov,exponential ltering and `1 norm based regularization methods.

Cette thèse concerne le diagnostic de processus dynamiques basée sur la Théorie des Observations Datées, une théorie mathématique conçue pour la modélisation et le raisonnement à partir de données datées. Les contributions présentées dans ce mémoire sont 1) une extension de la méthodologie d'ingénierie des connaissances TOM4D (Timed Observation Modeling for Diagnosis) aux réseaux de processus dynamiques, 2) l'algorithme temps réel et any-time TOM4E (Timed Observation Management for Explanation) qui utilise les modèles TOM4D pour diagnostiquer les comportements dans un réseau de processus dynamiques à partir de données datées et 3) l'application de TOM4D et TOM4E au diagnostic du barrage hydraulique des Sapins (France), un problème particulièrement difficile. TOM4D est une approche de diagnostic à partir de multiples modèles dirigée par la syntaxe ou l'introduction de la sémantique est contrôlée par la Combinaison de l'approche conceptuelle de CommonKADS au tétraèdre des états de la physique newtonienne. Les fonctions Detect, Describe et Explain de TOM4E utilisent les modèles d'observation déduit des modèles de comportement de TOM4D pour identifier les comportements potentiels des processus. Pour des raisons de simplicité, la présentation de TOM4D et de TOM4E est effectuée à l'aide d'un exemple didactique tirée de la littérature spécialisée dans le domaine du diagnostic. L'application au diagnostic du barrage des Sapins démontre l'intérêt de l'approche : leur usage aurait permis d'identifier le premier problème huit ans avant sa quasi-destruction, la présence d'eau étant mise en évidence sept ans avant. / This thesis proposes a diagnosis approach of dynamic process based on the Timed Observation Theory, a mathematical framework for modeling and reasoning about dynamic process from timed data. The contributions of this works are i) an extension of the TOM4D (Timed Observation Modeling for Diagnosis) Knowledge Engineering methodology to networks of dynamic processes, ii) a real-time and any-time diagnosis algorithm called TOM4E (Timed Observation Management for Explanation) that uses the TOM4D models to diagnose behaviors in a network of dynamic processes and iii) the application of TOM4D and TOM4E to the diagnosis of the French Sapin's hydraulic dam, a particularly difficult real-world diagnosis problem. TOM4D is a is a primarily syntax-driven approach of Multi-Model Based Diagnosis where semantic content is introduced in a gradual and controlled way through the combination of the CommonKADS conceptual approach and the Tetrahedron of States of Newton's physical laws.TOM4E algorithm is based on the Detect, Describe and Explain functions which uses observation models translated from the TOM4D behavioral models. For simplicity reasons, the presentation of TOM4D and TOM4E is made with a unique didactic example provided from the literature of the diagnosis domain. The example of Sapin's dam makes the demonstration of the interest of the proposed approach: using them, the first Sapin's dam problem would have been identified eight years before its quasi-failure, and the presence of water being highlighted seven years before.

L'hydrogène… Cet élément simple et très abondant pourrait être un contributeur clé à la transition énergétique, mais dans quelles conditions technico-économiques et politiques ? Cette thèse propose une contribution à l'évaluation de la faisabilité de pénétration de l'hydrogène dans le système énergétique, en mettant en oeuvre différents modèles qui permettent des éclairages complémentaires. Elle se concentre sur l’hydrogène bas carbone, obtenu par électrolyse de l’eau.Notre analyse multirégionale qui porte sur le contexte énergétique européen, américain, chinois et japonais (régions qui présentent des défis énergétiques contrastés) montre que les politiques énergétiques actuelles ne facilitent qu’une faible pénétration de l'hydrogène dans le système énergétique, lui permettant de réaliser environ 3% de l’effort à fournir par les quatre régions afin de limiter l’augmentation de la température à 2°C par rapport aux niveaux préindustriels. Nous soulignons dans cette thèse que l’injection d’hydrogène dans les réseaux de gaz naturel qui permet dans une certaine mesure d’éviter des fuites de méthane à fort pouvoir de réchauffement, pourrait jouer un rôle significatif dans la réalisation des objectifs de réduction des émissions de gaz à effet de serre.L'analyse des marchés de l'hydrogène a été menée en deux étapes. Tout d'abord, chaque marché (industriel ou énergétique) a été abordé individuellement afin d’établir des coûts d'entrée sur ce marché (pour les différents contextes énergétiques considérés). Ensuite, les différentes applications de l’hydrogène ont été resituées en interaction avec l’ensemble du système énergétique à travers le modèle TIMES-PT et un cas d’étude portant sur le Portugal, permettant ainsi d’examiner le potentiel de couplage entre les secteurs énergétiques rendu possible par l’hydrogène. Ces travaux ont permis de qualifier l'attractivité des différents marchés, celui de la mobilité apparaissant comme le plus favorable.Nous nous sommes ensuite intéressés aux coûts requis sur l'ensemble de la chaîne d'approvisionnement en hydrogène afin de pénétrer le marché de la mobilité.Pour ce faire, nous avons utilisé des modèles avec une maille géographique et temporelle fine (GLAES, EuroPower et InfraGis), en commençant par l’étape de production. Nous avons étudié le rôle potentiel de l'hydrogène pour la fourniture de flexibilité au système électrique dans un contexte de forte pénétration des énergies renouvelables intermittentes en France. Nos résultats montrent que l’hydrogène pourrait permettre non seulement d’éviter d’écrêter la production d’énergies renouvelables (entre 1,4 et 7,9 TWh en fonction du scénario de capacité d’interconnexion), mais pourrait aussi mettre à profit l’énergie nucléaire disponible (bas carbone donc), évitant par-là d’imposer de fortes rampes de puissances aux centrales. Cependant, une attention particulière doit être accordée au taux d'utilisation de l'électrolyseur afin de maintenir les coûts de production d'hydrogène suffisamment bas.Enfin, nous nous sommes concentrés sur l’approvisionnement de l’hydrogène, depuis les sites de production jusqu’à l’utilisation pour la mobilité, la question de l’infrastructure étant un problème majeur entravant les investissements dans l’hydrogène. Cinq filières d’approvisionnement (transport et distribution) ont été développées à la maille régionale et comparées sur le plan économique pour le cas français. Nos résultats montrent que, lors des toutes premières phases de pénétration du marché (scénario 1%), il est plus intéressant de privilégier la production décentralisée. / Hydrogen… This simple, very abundant element holds great promise to contribute to the transition towards a cleaner future energy system, but under which techno-economic and political conditions? This thesis is a contribution to the assessment of the hydrogen penetration feasibility into the energy system, using a multi-model approach. The focus is put on low-carbon hydrogen, obtained by electrolysis.Our multi-regional analysis on the European, American, Chinese and Japanese energy context (presenting contrasted energy challenges) show that, with the current energy policies implemented which result in a modest penetration of hydrogen into the energy system, hydrogen may achieve approximately 3% of the effort that needs to be done by the four regions, in order to limit the increase of the temperature to 2°C, compared to preindustrial levels. We highlight in this thesis that blending hydrogen with natural gas, and thereby avoiding methane leakages to a certain extent, may represent a significant contribution in achieving the carbon mitigation goals.The hydrogen market analysis has been carried out following two steps. First, each market (industrial and energy-related) was tackled aside in order to propose market entry costs considering the four energy contexts and investigate the timeframe of the market penetration potential. Then, the different hydrogen applications were examined within the overall energy system through the TIMES-PT model (for a Portugal case study), allowing to investigate the hydrogen potential for energy sector coupling. Based on this work, the markets attractiveness was evaluated: mobility (using fuel cell vehicles) appears to be the most favourable.Then, we tackled the required costs over the whole hydrogen supply chain in order to enter the mobility market.To do so, we used temporally and spatially resolved models (GLAES, EuroPower and InfraGis) starting with the production side where we studied the hydrogen potential role in providing the electricity system with flexibility and the impact of such electrolysis operation on the hydrogen generation costs in the context of high shares of renewable energies in France. Our results show that hydrogen can contribute to improve the flexibility of the electric system by allowing avoiding renewable curtailment (between 1.4 and 7.9 TWh depending on the interconnection capacity scenario) but also by taking advantage of nuclear plant available energy (thereby avoiding nuclear ramping), the latter ensuring a low carbon and low cost electricity provision. However, a special attention needs to be dedicated to the utilisation rate of the electrolyser, to keep the hydrogen production costs low enough.Last but not least, we focused on how to link the hydrogen production sites and its final use for mobility applications, the delivery infrastructure being a major issue hampering the hydrogen investments. Five transport and delivery pathways were geographically designed and economically assessed, for the French case. According to our findings, during the very first market penetration phases (1% scenario), it is more interesting to start with decentralised production that proved to be less expensive for the whole pathway at this stage.

This study conducted an experiment that compares the multi-model NoSQL DBMS ArangoDB with other NoSQL DBMS, in terms of the average response time of queries. The DBMS compared in this experiment are the following: Redis, MongoDB, Couchbase, and OrientDB. The hypothesis that is answered in this study is the following: “There is a significant difference between ArangoDB, OrientDB, Couchbase, Redis, MongoDB in terms of the average response time of queries”. This is examined by comparing the average response time of 1 000, 100 000, and 1 000 000 queries between these database systems. The results show that ArangoDB performs worse compared to the other DBMS. Examples of future work include using additional DBMS in the same experiment and replacing ArangoDB with another multi-model DBMS to decide whether such a DBMS, in general, performs worse than single-model DBMS.

No description available.

With the widespread use of the Internet, the popularity of web applications has significantly increased. Such applications are subject to unpredictable workload conditions that vary from time to time. For example, an e-commerce website may face higher workloads than normal during festivals or promotional schemes. Such applications are critical and performance related issues, or service disruption can result in financial losses. Cloud computing with its attractive feature of dynamic resource provisioning (elasticity) is a perfect match to host such applications. The rapid growth in the usage of cloud computing model, as well as the rise in complexity of the web applications poses new challenges regarding the effective monitoring and management of the underlying cloud computational resources. This thesis investigates the state-of-the-art elastic methods including the models and techniques for the dynamic management and provisioning of cloud resources from a service provider perspective. An elastic controller is responsible to determine the optimal number of cloud resources, required at a particular time to achieve the desired performance demands. Researchers and practitioners have proposed many elastic controllers using versatile techniques ranging from simple if-then-else based rules to sophisticated optimisation, control theory and machine learning based methods. However, despite an extensive range of existing elasticity research, the aim of implementing an efficient scaling technique that satisfies the actual demands is still a challenge to achieve. There exist many issues that have not received much attention from a holistic point of view. Some of these issues include: 1) the lack of adaptability and static scaling behaviour whilst considering completely fixed approaches; 2) the burden of additional computational overhead, the inability to cope with the sudden changes in the workload behaviour and the preference of adaptability over reliability at runtime whilst considering the fully dynamic approaches; and 3) the lack of considering uncertainty aspects while designing auto-scaling solutions. This thesis seeks solutions to address these issues altogether using an integrated approach. Moreover, this thesis aims at the provision of qualitative elasticity rules. This thesis proposes a novel biologically-inspired switched feedback control methodology to address the horizontal elasticity problem. The switched methodology utilises multiple controllers simultaneously, whereas the selection of a suitable controller is realised using an intelligent switching mechanism. Each controller itself depicts a different elasticity policy that can be designed using the principles of fixed gain feedback controller approach. The switching mechanism is implemented using a fuzzy system that determines a suitable controller/- policy at runtime based on the current behaviour of the system. Furthermore, to improve the possibility of bumpless transitions and to avoid the oscillatory behaviour, which is a problem commonly associated with switching based control methodologies, this thesis proposes an alternative soft switching approach. This soft switching approach incorporates a biologically-inspired Basal Ganglia based computational model of action selection. In addition, this thesis formulates the problem of designing the membership functions of the switching mechanism as a multi-objective optimisation problem. The key purpose behind this formulation is to obtain the near optimal (or to fine tune) parameter settings for the membership functions of the fuzzy control system in the absence of domain experts’ knowledge. This problem is addressed by using two different techniques including the commonly used Genetic Algorithm and an alternative less known economic approach called the Taguchi method. Lastly, we identify seven different kinds of real workload patterns, each of which reflects a different set of applications. Six real and one synthetic HTTP traces, one for each pattern, are further identified and utilised to evaluate the performance of the proposed methods against the state-of-the-art approaches.

Made available in DSpace on 2014-07-29T14:57:52Z (GMT). No. of bitstreams: 1 dissertacao Fabiana Mendes.pdf: 748371 bytes, checksum: 5be0381086e13f51513dc163e7b41170 (MD5) Previous issue date: 2010-04-16 / To use just one quality model in a Information Technology Process Improvement (ITPI) iniciative means just one: language, thought about how to fit organization needs, and way to conduct improvements and appraisals. Despite, to use just one model also means it could exist other ways to deal with ITIPI needs, but this is not going to be considered. To use more than one model, hence, means to model a best solution. Mono-Model ITPI initiatives have similar nature if compared with multi-model initiatives. However, there are multi-model specific challenges like: integration of improvement initiatives, of appraisals, and of quality model. Despite of being possible to model better solutions through multi model ITPI, there are little works about it. This research review works related to this subject, presents challenges multi-model list, and comes up with an approach to deal with these challenges. This approach has the advantage to be independent of the conduct method of improvement, quality model or project improvement phase. / Utilizar um único modelo de qualidade em uma iniciativa de Melhoria de Processos de TI (MPTI) em uma organização significa também única: linguagem, maneira de pensar sobre como atender as necessidades de MPTI, e maneira de conduzir as melhorias e as avaliações. Apesar disso, também significa que poderiam existir outras formas de lidar com estas necessidades, mas que isso sequer será considerado. Utilizar mais de um modelo, portanto, significa que se pode modelar uma melhor solução para os problemas detectados. Iniciativas de MPTI baseadas em um único modelo possuem natureza semelhante àquelas que utilizam múltiplos modelos. Entretanto, podem ser visualizados desafios específicos nas iniciativas multi-modelo, tais como a integração de iniciativas de melhoria, de avaliações, e de modelos de qualidade. Apesar de ser possível modelar melhores soluções através de iniciativas de MPTI multi-modelo, poucos trabalhos tratam de MPTI multi-modelo. Esta pesquisa, além de revisar trabalhos relacionados ao tema, apresenta como resultado uma lista de desafios encontrados em iniciativas multi-modelo e propõe uma abordagem para lidar com eles. Esta abordagem possui como vantagem, em relação às outras analisadas neste trabalho, o fato de ser independente do método de implantação de MPTI que tem sido utilizado na organização, do modelo de qualidade empregado e da fase em que o projeto de melhoria se encontra.

La réalisation de tâches de manipulation dextres requiert une complexité aussi bien dans la conception de préhenseur robotique que dans la synthèse de leurs lois de commande. Une optimisation de la mécatronique de ces systèmes permet de répondre aux contraintes d'intégration fonctionnelle en se passant de capteurs de force terminaux. L'utilisation de mécanismes réversibles rend alors possible la détermination du positionnement du système dans l'espace libre et la détection de son interaction avec les objets manipulés, grâce aux mesures proprioceptives inhérentes aux actionneurs électriques. L'objectif de cette thèse est de parvenir synthétiser, dans le contexte articulaire (un degré-de-liberté), une commande adaptée à la manipulation en tenant compte de ces particularités mécaniques. La méthode proposée est basée sur une commande robuste par rapport aux non-linéarités structurelles dues aux effets gravitationnels et aux frottements secs d'une part et par rapport aux rigidités variables des objets manipulés. L'approche choisie nécessite la connaissance précise de la configuration du système étudié à chaque instant. Une représentation dynamique de son comportement permet de synthétiser un capteur logiciel pour l'estimation des grandeurs indispensables à la commande. Ces différentes étapes sont validées par des essais expérimentaux pour justifier la démarche choisie menant à une commande adaptée à la manipulation d'objets. / The realization of dexterous manipulation tasks requires a complexity in robotic hands design as well as in their control laws synthesis. A mecatronical optimization of these systems helps to answer for functional integration constraints by avoiding external force sensors. Back-drivable mechanics allows the free-space positioning determination of such system as far as the detection of its interaction with a manipulated object thanks to proprioceptives measures at electric actuator level. The objective of this thesis is to synthesize a control law adapted to object manipulation by taking into account these mechanical properties in a one degree-of-freedom case. The proposed method is based on a robust control according to structural non-linearities due to gravitational effects and dry frictions on the one hand and with regard to a variable rigidity of manipulated objects on the other hand. The chosen approach requires a precise knowledge of the system configuration at all time. A dynamic representation of its behavior enables a software sensor synthesis for the exteroceptives variables estimation in a control law application purpose. The different steps are experimentally validated in order to justify the chosen approach leading to object manipulation.