A Modelagem Orientada a Objetos (MOO) é o processo de construção de modelos de sistemas através da identificação e definição de um conjunto de objetos relacionados, que comportam-se e colaboram entre si conforme os requisitos estabelecidos para o sistema. Esta definição inclui os três aspectos ortogonais, ou dimensões, deste tipo de modelagem: a dimensão estrutural dos objetos, a dimensão dinâmica do comportamento e a dimensão funcional dos requisitos. Conforme a importância relativa dada a cada uma destas dimensões, podem ser definidas três estratégias possíveis para conduzir a MOO. Estas estratégias são as dirigidas por dados, por comportamento e por processos. A estratégia dirigida por processos já esta superada. Atualmente, a estratégia dirigida por dados domina na maioria das técnicas de MOO. A estratégia dirigida por comportamento propõe que a estrutura dos objetos em um sistema pode ser determinada a partir do comportamento externo e interno que o sistema deve apresentar. Esta idéia é interessante, porque permite introduzir tardiamente o encapsulamento na MOO. Conforme é argumentado neste trabalho, as vantagens atribuídas a orientação a objetos são de implementação, isto é, a decisão de orientar ou não a objetos é, na realidade, uma decisão de design. Ao introduzir o encapsulamento na modelagem inicial do sistema, ganha-se o benefício da continuidade estrutural ao custo de colocar a MOO mais perto do design. Neste contexto, este trabalho apresenta um processo de modelagem conceitual de sistemas do ponto de vista comportamental que introduz tardiamente o encapsulamento da orientação a objetos como primeiro passo de design. Em outras palavras, é proposta uma técnica de modelagem sob uma estratégia dirigida por comportamento (privilegiando, assim, o aspecto dinâmico dos sistemas) com o suficiente poder de expressão para, ao mesmo tempo, permitir a modelagem de sistemas de informação no nível conceitual e derivar dos modelos dinâmicos obtidos uma representação estrutural orientada a objetos. 0 sistema, na concepção desta proposta, é composto por um conjunto de processos concorrentes, cada um dos quais recebe um estimulo do ambiente, realiza um tratamento especifico sobre ele e gera para o ambiente uma resposta. Os estímulos externos são decompostos em conjuntos de eventos concorrentes tratados no interior do processo. As ações realizadas no interior do mesmo são compostas nas respostas geradas para o exterior. Os processos são modelados comportamentalmente, utilizando o formalismo proposto High-Level Statecharts (HLS). HLS é uma extensão dos statecharts de Harel. As principais extensões propostas são a introdução de estados "parametrizados" usando variáveis e a representação genérica de conjuntos de estados concorrentes e exclusivos. 0 modelo de processos e desintegrado em unidades de comportamento que tratam das mesmas variáveis. Estas unidades são integradas em um modelo de ciclos de vida para estas variáveis. Finalmente, apos a aplicação da técnica de modelagem conceitual, e obtido um modelo estrutural orientado a objetos. Este modelo e derivado utilizando unicamente informações contidas nos modelos dinâmicos gerados no processo da técnica proposta. No modelo estrutural são identificadas classes, objetos, atributos, associações estáticas, hierarquias de herança e operações. Todo o processo e exemplificado utilizando o problema padrão de preparação de congressos da IFIP. / Object-Oriented Modeling (OOM) is the process of construction of systems models, through an identification and definition of a set of relating objects. These objects have a collaborative behavior according to the system requirements previously defined. This definition includes three modeling aspects or dimensions: object structural dimension, behavior dynamic dimension and requirements functional dimension. Depending on a relative importance of each dimension, three possible strategies to drive OOM are defined. The strategies are: data-driven, behavior-driven and process-driven. Process-driven strategy is obsolete. Nowadays, data-driven is the dominant strategy in the world of OOM techniques. Behavior-driven strategy suggests both internal and external system behaviors define its object structure. This idea is attractive because it allows a late encapsulation in the OOM. As explained in this work, the main advantage to use object-orientation is for implementation. So, to object-orient or not to object-orient is a design decision. If encapsulation is introduced in the very beginning of systems modeling, the structural continuity is achieved at the cost of pulling OOM closer to design. In this context, the work presents a process of systems conceptual modeling using a behavioral point of view. This process introduces object-oriented encapsulation lately as a first step in the design phase. In other words, this work is a proposal of a modeling technique under a behavior-driven strategy (focusing the dynamic aspect of the systems) with enough expression power to model information systems at conceptual level and, at the same time, to derive of an object-oriented structural representation from the dynamic models. As conceived in the proposal, a system is composed by a set of concurrent processes. Each process receives a stimuli from the environment, makes a specific treatment on it and generates a response to the environment. The external stimuli is decomposed into a set of concurrent events which are internally handled by the process. Actions internally performed by the process are composed into a response which is sent outside the process. Processes are behaviorally modeled using a proposed formalism called High-Level Statecharts (HLS). HLS is a extension of Harel's statecharts. The main extensions proposed are parameterized states using variables and generic representation of concurrent and exclusive sets of states. Process model is disintegrated into behavior units handling the same variables. The units are integrated into a life cycle model for these variables. Finally, after the modeling technique has been applied, an object-oriented structural model is obtained. This model is derived exclusively using information from the dynamic models constructed during the modeling process. Classes, objects, attributes, static associations, inheritance hierarchies and operations in the structural model are identified. Examples used in all the modeling process are taken from the standard problem of IFIP conference.

Cette thèse étudie les problèmes de diagnostic itératif et propose des outilsd’aide au diagnostic interactif. Différents processus de diagnostic où des outils interactifs homme-automate sont utiles, sont présentés. Ces outils permettent de résoudre des difficultés liées à la représentation d’un grand nombre d’éléments d’un système, des difficultés liées à la représentation du comportement et du fonctionnement d’un système et des difficultés liées à l’explicitation de l’expertise. Nos travaux ont conduit à la conception de différents types d’outils interactifs d’aide au diagnostic. Le premier permet d’exploiter des représentations structuro-fonctionnelles pour construire et résoudre progressivement un problème de diagnostic.Le second outil interactif permet d’exploiter des modèles de comportement construit au fur et à mesure de la résolution d’un problème de diagnostic. Enfin, un dernier outil a étéproposé pour montrer qu’il est possible de prendre compte la connaissance implicite d’un expert dans la résolution de problème de diagnostic. Un problème de diagnostic est donc présenté comme un processus itératif avec des interactions homme-automate. / This PhD thesis studies the iterative diagnosis problems and provides thecomputer-aided diagnostic tool for interactive diagnosis. Different diagnosis processes wherethe tool to support human-machine interaction are useful, are presented. These tools help totackle difficulties related to the representation of a large number of elements in a system,difficulties related to the representation of the behavior functioning of a system and difficultiesencountered while expliciting the expertise. Our work led to the design of different interactivetools to support the diagnosis process. The first tool allows to exploit the structural-functionalmodeling to build and solve progressively a diagnosis problem. The second interactive toolallows to exploit the behavioral models built step by step in the diagnosis process and tosolve the diagnosis problem. The final tool was proposed to show that it is possible to takeinto account the implicit knowledge of an expert in order to solve the diagnosis problem.A diagnosis problem is therefore presented as an iterative process with human-machineinteractions.

A Modelagem Orientada a Objetos (MOO) é o processo de construção de modelos de sistemas através da identificação e definição de um conjunto de objetos relacionados, que comportam-se e colaboram entre si conforme os requisitos estabelecidos para o sistema. Esta definição inclui os três aspectos ortogonais, ou dimensões, deste tipo de modelagem: a dimensão estrutural dos objetos, a dimensão dinâmica do comportamento e a dimensão funcional dos requisitos. Conforme a importância relativa dada a cada uma destas dimensões, podem ser definidas três estratégias possíveis para conduzir a MOO. Estas estratégias são as dirigidas por dados, por comportamento e por processos. A estratégia dirigida por processos já esta superada. Atualmente, a estratégia dirigida por dados domina na maioria das técnicas de MOO. A estratégia dirigida por comportamento propõe que a estrutura dos objetos em um sistema pode ser determinada a partir do comportamento externo e interno que o sistema deve apresentar. Esta idéia é interessante, porque permite introduzir tardiamente o encapsulamento na MOO. Conforme é argumentado neste trabalho, as vantagens atribuídas a orientação a objetos são de implementação, isto é, a decisão de orientar ou não a objetos é, na realidade, uma decisão de design. Ao introduzir o encapsulamento na modelagem inicial do sistema, ganha-se o benefício da continuidade estrutural ao custo de colocar a MOO mais perto do design. Neste contexto, este trabalho apresenta um processo de modelagem conceitual de sistemas do ponto de vista comportamental que introduz tardiamente o encapsulamento da orientação a objetos como primeiro passo de design. Em outras palavras, é proposta uma técnica de modelagem sob uma estratégia dirigida por comportamento (privilegiando, assim, o aspecto dinâmico dos sistemas) com o suficiente poder de expressão para, ao mesmo tempo, permitir a modelagem de sistemas de informação no nível conceitual e derivar dos modelos dinâmicos obtidos uma representação estrutural orientada a objetos. 0 sistema, na concepção desta proposta, é composto por um conjunto de processos concorrentes, cada um dos quais recebe um estimulo do ambiente, realiza um tratamento especifico sobre ele e gera para o ambiente uma resposta. Os estímulos externos são decompostos em conjuntos de eventos concorrentes tratados no interior do processo. As ações realizadas no interior do mesmo são compostas nas respostas geradas para o exterior. Os processos são modelados comportamentalmente, utilizando o formalismo proposto High-Level Statecharts (HLS). HLS é uma extensão dos statecharts de Harel. As principais extensões propostas são a introdução de estados "parametrizados" usando variáveis e a representação genérica de conjuntos de estados concorrentes e exclusivos. 0 modelo de processos e desintegrado em unidades de comportamento que tratam das mesmas variáveis. Estas unidades são integradas em um modelo de ciclos de vida para estas variáveis. Finalmente, apos a aplicação da técnica de modelagem conceitual, e obtido um modelo estrutural orientado a objetos. Este modelo e derivado utilizando unicamente informações contidas nos modelos dinâmicos gerados no processo da técnica proposta. No modelo estrutural são identificadas classes, objetos, atributos, associações estáticas, hierarquias de herança e operações. Todo o processo e exemplificado utilizando o problema padrão de preparação de congressos da IFIP. / Object-Oriented Modeling (OOM) is the process of construction of systems models, through an identification and definition of a set of relating objects. These objects have a collaborative behavior according to the system requirements previously defined. This definition includes three modeling aspects or dimensions: object structural dimension, behavior dynamic dimension and requirements functional dimension. Depending on a relative importance of each dimension, three possible strategies to drive OOM are defined. The strategies are: data-driven, behavior-driven and process-driven. Process-driven strategy is obsolete. Nowadays, data-driven is the dominant strategy in the world of OOM techniques. Behavior-driven strategy suggests both internal and external system behaviors define its object structure. This idea is attractive because it allows a late encapsulation in the OOM. As explained in this work, the main advantage to use object-orientation is for implementation. So, to object-orient or not to object-orient is a design decision. If encapsulation is introduced in the very beginning of systems modeling, the structural continuity is achieved at the cost of pulling OOM closer to design. In this context, the work presents a process of systems conceptual modeling using a behavioral point of view. This process introduces object-oriented encapsulation lately as a first step in the design phase. In other words, this work is a proposal of a modeling technique under a behavior-driven strategy (focusing the dynamic aspect of the systems) with enough expression power to model information systems at conceptual level and, at the same time, to derive of an object-oriented structural representation from the dynamic models. As conceived in the proposal, a system is composed by a set of concurrent processes. Each process receives a stimuli from the environment, makes a specific treatment on it and generates a response to the environment. The external stimuli is decomposed into a set of concurrent events which are internally handled by the process. Actions internally performed by the process are composed into a response which is sent outside the process. Processes are behaviorally modeled using a proposed formalism called High-Level Statecharts (HLS). HLS is a extension of Harel's statecharts. The main extensions proposed are parameterized states using variables and generic representation of concurrent and exclusive sets of states. Process model is disintegrated into behavior units handling the same variables. The units are integrated into a life cycle model for these variables. Finally, after the modeling technique has been applied, an object-oriented structural model is obtained. This model is derived exclusively using information from the dynamic models constructed during the modeling process. Classes, objects, attributes, static associations, inheritance hierarchies and operations in the structural model are identified. Examples used in all the modeling process are taken from the standard problem of IFIP conference.

Ce manuscrit propose une approche méthodologique pour la constitution d’une chaîne complète de vidéosurveillance pour des environnements naturellement encombrés. Nous identifions et levons un certain nombre de verrous méthodologiques et technologiques inhérents : 1) à l’acquisition de séquences vidéo en milieu naturel, 2) au traitement d’images, 3) au suivi multi-cibles, 4) à la découverte et la modélisation de motifs comportementaux récurrents, et 5) à la fusion de données. Le contexte applicatif de nos travaux est la surveillance apicole, et en particulier, l’étude des trajectoires des abeilles en vol devant la ruche. De ce fait, cette thèse se présente également comme une étude de faisabilité et de prototypage dans le cadre des deux projets interdisciplinaires EPERAS et RISQAPI (projets menées en collaboration avec l’INRA Magneraud et le Muséum National d’Histoire Naturelle). Il s’agit pour nous informaticiens et pour les biologistes qui nous ont accompagnés, d’un domaine d’investigation totalement nouveau, pour lequel les connaissances métiers, généralement essentielles à ce genre d’applications, restent encore à définir. Contrairement aux approches existantes de suivi d’insectes, nous proposons de nous attaquer au problème dans l’espace à trois dimensions grâce à l’utilisation d’une caméra stéréovision haute fréquence. Dans ce contexte, nous détaillons notre nouvelle méthode de détection de cibles appelée segmentation HIDS. Concernant le calcul des trajectoires, nous explorons plusieurs approches de suivi de cibles, s’appuyant sur plus ou moins d’a priori, susceptibles de supporter les conditions extrêmes de l’application (e.g. cibles nombreuses, de petite taille, présentant un mouvement chaotique). Une fois les trajectoires collectées, nous les organisons selon une structure de données hiérarchique et mettons en œuvre une approche Bayésienne non-paramétrique pour la découverte de comportements émergents au sein de la colonie d’insectes. L’analyse exploratoire des trajectoires issues de la scène encombrée s’effectue par classification non supervisée, simultanément sur des niveaux sémantiques différents, et où le nombre de clusters pour chaque niveau n’est pas défini a priori mais est estimé à partir des données. Cette approche est dans un premier temps validée à l’aide d’une pseudo-vérité terrain générée par un Système Multi-Agents, puis dans un deuxième temps appliquée sur des données réelles. / This manuscript provides the basis for a complete chain of videosurveillence for naturally cluttered environments. In the latter, we identify and solve the wide spectrum of methodological and technological barriers inherent to : 1) the acquisition of video sequences in natural conditions, 2) the image processing problems, 3) the multi-target tracking ambiguities, 4) the discovery and the modeling of recurring behavioral patterns, and 5) the data fusion. The application context of our work is the monitoring of honeybees, and in particular the study of the trajectories bees in flight in front of their hive. In fact, this thesis is part a feasibility and prototyping study carried by the two interdisciplinary projects EPERAS and RISQAPI (projects undertaken in collaboration with INRA institute and the French National Museum of Natural History). It is for us, computer scientists, and for biologists who accompanied us, a completely new area of investigation for which the scientific knowledge, usually essential for such applications, are still in their infancy. Unlike existing approaches for monitoring insects, we propose to tackle the problem in the three-dimensional space through the use of a high frequency stereo camera. In this context, we detail our new target detection method which we called HIDS segmentation. Concerning the computation of trajectories, we explored several tracking approaches, relying on more or less a priori, which are able to deal with the extreme conditions of the application (e.g. many targets, small in size, following chaotic movements). Once the trajectories are collected, we organize them according to a given hierarchical data structure and apply a Bayesian nonparametric approach for discovering emergent behaviors within the colony of insects. The exploratory analysis of the trajectories generated by the crowded scene is performed following an unsupervised classification method simultaneously over different levels of semantic, and where the number of clusters for each level is not defined a priori, but rather estimated from the data only. This approach is has been validated thanks to a ground truth generated by a Multi-Agent System. Then we tested it in the context of real data.

Behavioral animation has become popular for creating virtual characters that are autonomous agents and thus self-animating. This is useful for lessening the workload of human animators, populating virtual environments with interactive agents, etc. Unfortunately, current behavioral animation techniques suffer from three key problems: (1) deliberative behavioral models (i.e., cognitive models) are slow to execute; (2) interactive virtual characters cannot adapt online due to interaction with a human user; (3) programming of behavioral models is a difficult and time-intensive process. This dissertation presents a collection of papers that seek to overcome each of these problems. Specifically, these issues are alleviated through novel machine learning schemes. Problem 1 is addressed by using fast regression techniques to quickly approximate a cognitive model. Problem 2 is addressed by a novel multi-level technique composed of custom machine learning methods to gather salient knowledge with which to guide decision making. Finally, Problem 3 is addressed through programming-by-demonstration, allowing a non technical user to quickly and intuitively specify agent behavior.

L’objet de la thèse est la modélisation de systèmes hétérogènes intégrant différents domaines de la physique et à signaux mixtes, numériques et analogiques (AMS). Une étude approfondie de différentes techniques d’extraction et de calibration de modèles comportementaux de composants analogiques à différents niveaux d’abstraction et de précision est présentée. Cette étude a mis en lumière trois approches principales qui ont été validées par la modélisation de plusieurs applications issues de divers domaines: un amplificateur faible bruit (LNA), un capteur chimique basé sur des ondes acoustiques de surface (SAW), le développement à plusieurs niveaux d’abstraction d’un capteur CMOS vidéo, et son intégration dans une plateforme industrielle. Les outils développés sont basés sur les extensions AMS du standard IEEE 1666 SystemC mais les techniques proposées sont facilement transposables à d’autres langages tels que VHDL-AMS ou Verilog-AMS utilisés en conception de dispositifs mixtes. / The thesis objective is the modeling of heterogeneous systems. Such systems integrate different physical domains (mechanical, chemical, optical or magnetic) therefore integrate analog and mixed- signal (AMS) parts. The aim is to provide a methodology based on high-level modeling for assisting both the design and the verification of AMS systems. A study on different techniques for extracting behavioral models of analog devices at different abstraction levels and computational weights is presented. Three approaches are identified and regrouped in three techniques. These techniques have been validated through the virtual prototyping of different applications issued from different domains: a low noise amplifier (LNA), a surface acoustic wave-based (SAW) chemical sensor, a CMOS video sensor with models developed at different abstraction levels and their integration within an industrial platform. The flows developed are based on the AMS extensions of the SystemC (IEEE 1666) standard but the methodologies can be implemented using other Analog Hardware Description Languages (VHDL-AMS, Verilog-AMS) typically used for mixed-signal microelectronics design.