Em sistemas multiagentes (SMA), a impossibilidade de controle do comportamento interno dos agentes propicia a emergência de dilemas sociais. Uma das possíveis abordagens para resolver tais dilemas é a abordagem social, que tem no conceito de reputação seu principal mecanismo de controle. Para agilizar a obtenção da reputação, torna-se essencial que os agentes cooperem, trocando informações a respeito de reputação. Assim, a partir de diversas observações, avaliações e troca de informações sobre reputação pode-se observar o surgimento da ordem social no sistema global. Entretanto, a heterogeneidade semântica entre os diferentes modelos de reputação torna a interação de agentes um problema, principalmente em SMA abertos. Este trabalho propõe uma arquitetura baseada em serviços, denominada SOARI, para o tratamento deste problema. Esta arquitetura apóia o mapeamento e tradução de conceitos entre ontologias de modelos de reputação distintos através do uso de uma ontologia comum no domínio da reputação. Para ilustrar o uso da arquitetura, foram realizados diversos experimentos nos domínios de comércio eletrônico e de avaliação de obras de arte. Neste último caso, analisou-se os impactos causados nos agentes como conseqüência da habilitação de uma maior expressividade e heterogeneidade na comunicação sobre reputação. / In Multiagent Systems (MAS), controlling the internal agent\'s behavior is impossible and therefore social dilemma may emerge. One possible approach to solve the problem is the social approach, which considers reputation as its main control mechanism. In order to accelerate the reputation evaluation, it is essential that the cooperating agents could exchange information about reputation. Thus, by using observations, evaluations and information exchange about reputation the social order in the global system may occur. However, the semantic heterogeneity of different reputation models makes agent interaction a problem, especially in open MAS. This work proposes a service oriented architecture, named SOARI, to deal with this problem. This architecture supports concept mapping and translation among different reputation model ontologies by using a common ontology. In order to illustrate its deployment, several experiments in the electronic commerce and art appraisal domains were performed. In the latter it was analyzed the effects caused on the agents as a consequence of a more expressive and heterogeneous communication about reputation.

O objetivo deste trabalho é o desenvolvimento de uma ferramenta de simulação baseada em multiagentes para calcular a emergência de parcerias entre múltiplos agentes. A metodologia adotada foi tomar como base o sistema PART-NET [CONTE’98], que efetua o cálculo da emergência de parcerias entre uma dupla de agentes, e alterá-lo de acordo com o objetivo proposto. Dessa maneira, dividiu-se o trabalho em duas etapas. Na primeira, o sistema original foi reescrito na linguagem Java, promovendo benefícios operacionais, como a melhora da interface e a apresentação dos resultados de forma gráfica. Esse sistema intermediário foi denominado PartNET+. A segunda etapa constituiu-se em expandir a funcionalidade do sistema intermediário, tornando sua arquitetura de agentes mais completa pela adição de planos, o que permite novos tipos de parcerias. Para processar essas parcerias, criou-se um novo algoritmo para cálculo de parcerias entre múltiplos agentes, com base em hipergrafos. O sistema resultante, que compreende estas extensões funcionais, foi denominado PartNET++. / The purpose of this work is the development of a multiagent based simulation tool to account partnership formation among multiple agents. In order to achieve that, the PART-NET [CONTE’98] system was adopted as a base. This system calculates the partnership formation of pairs of agents, and was altered according to the proposed objective. In this manner, the work was divided in two stages. In the first, the original system was redesigned using the Java language, bringing operational improvements to the predecessor, such as user interface enhancement and graphical display of the results. This intermediary system was named PartNET+. The second stage promotes new functionality to this intermediary system, adding plans to its agent architecture to allow new kinds of partnerships. A new algorithm for multiple agents partnership formation was developed based on hipergraphs, and the final system that handles this enhanced partnership formation was called PartNET++

Decision-related activities, such as bottom-up and top-down policy development, analysis, and planning, stand to benefit from the development and application of computer-based models that are capable of representing spatiotemporal social human behavior in local contexts. This is especially the case with our efforts to understand and search for ways to mitigate the context-specific effects of climate change, in which case such models need to include interacting social and ecological components. The development and application of such models has been significantly hindered by the challenges in designing artificial agents whose behavior is grounded in both empirical evidence and theory and in testing the ability of artificial agents to represent the behavior of real-world decision-makers. This dissertation advances our ability to develop such models by overcoming these challenges through the creation of: (a) three new frameworks, (b) two new methods, and (c) two new open-source modeling tools. The three new frameworks include: (a) the SOSIEL framework, which provides a theoretically-grounded blueprint for the development of a new generation of cognitive, multi-agent, and knowledge-based models that consist of agents empowered with cognitive architectures; (b) a new framework for analyzing the bounded rationality of decision-makers, which offers insight into and facilitates the analysis of the relationship between a decision situation and a decision-maker's decision; and (c) a new framework for analyzing the doubly-bounded rationality (DBR) of artificial agents, which does the same for the relationship between a decision situation and an artificial agent's decision. The two new methods include: (a) the SOSIEL method for acquiring and operationalizing decision-making knowledge, which advances our ability to acquire, process, and represent decision-making knowledge for cognitive, multi-agent, and knowledge-based models; and (b) the DBR method for testing the ability of artificial agents to represent human decision-making. The two open-source modeling tools include: (a) the SOSIEL platform, which is a cognitive, multi-agent, and knowledge-based platform for simulating human decision-making; and (b) an application of the platform as the SOSIEL Human Extension (SHE) to an existing forest-climate change model, called LANDIS-II, allowing for the analysis of co-evolutionary human-forest-climate interactions. To provide a context for examples and also guidelines for knowledge acquisition, the dissertation includes a case study of social-ecological interactions in an area of the Ukrainian Carpathians where LANDIS-II with SHE are currently being applied. As a result, this dissertation advances science by: (a) providing a theoretical foundation for and demonstrating the implementation of a next generation of models that are cognitive, multi-agent, and knowledge-based; and (b) providing a new perspective for understanding, analyzing, and testing the ability of artificial agents to represent human decision-making that is rooted in psychology.

Traffic simulation has become an invaluable part of the traffic engineering toolbox. However, the majority of driver models are designed to recreate traffic performance based on interactions among vehicles. In keeping with this pursuit, most are fundamentally built to avoid collisions. This limits the applicability of using these models for addressing safety concerns, especially those regarding pedestrian safety performance. However, by explicitly including some of the sources of human error, these limitations can, in theory, be overcome. While much work has been done toward including these human factors in simulation platforms, one key aspect of human behavior has been largely ignored: driver distraction. This work presents a novel approach to inclusion of driver distraction in a microsimulation or agent-based model. Distributions of distraction events and inter-distraction periods are derived from eye-glance data collected during naturalistic driving studies. The developed model of distraction is implemented -- along with perception errors, visual obstructions, and driver reaction times -- in a simulated mid-block pedestrian crossing. The results of this simulation demonstrate that excluding any of these human factors from the implemented driver model significantly alters conflict rates observed in the simulation. This finding suggests that inclusion of human factors is important in any microsimulation platforms used to analyze pedestrian safety performance.

En milieu urbain et périurbain, les transports et la mobilité sont étroitement liés aux activités socio-économiques. Ils sont influencés par la psychologie personnelle de l'individu, les normes sociales, la résistance au changement, l'attitude personnelle, les habitudes, les peurs et les croyances. A Beyrouth, la possession de véhicules a augmenté rapidement en raison de la pénurie de transports en commun amenant l’opportunité de développer des services de mobilité durable et d’accessibilité. Cela nécessite que les modes de transport soient abordables, efficaces et respectueux de l'environnement, ainsi que des technologies qui minimisent la consommation d'énergie en mettant l'accent sur trois points principaux: réduire les besoins de déplacement, adopter des modes de transport plus respectueux de l'environnement et améliorer l'efficacité de la technologie des véhicules. Cette thèse porte sur la modélisation orientée-agent du comportement de déplacement. Nous avons analysé les données obtenues à partir d’une enquête menée à Beyrouth. Une relation entre l’emploi, le statut social et le choix du mode de transport a été mise en évidence. Un modèle de simulation orientée agent, incluant les architectures et les algorithmes des agents, du trafic routier utilisant une approche économique institutionnelle a été proposé. Les questions scientifiques concernent les effets de facteurs économiques, sociaux, professionnels et liés aux ménages sur le choix d'un mode de transport. Ce travail fait partie d'un projet de la société S&A (Belgique et Liban) visant à fournir des outils permettant la transition des moyens de transport vers une pratique plus respectueuse. / In urban and peri-urban environments, transport and mobility are strongly linked to socio-economic activities. They are influenced by individual's personal psychology, social norms, resistance to change, personal attitude, habits, fears and beliefs. Beirut is a mono-centric city with transport supply poverty. Vehicle ownership has increased rapidly in response to the shortage of public transportation. Currently, there is an opportunity to develop mobility management services and establish sustainable mobility and accessibility in Beirut. This requires that transport modes be affordable, efficient and environmentally friendly, as well as technologies that minimize the energy consumption by focusing on three main points: reducing travel needs, adopting more environmentally friendly modes of transportation, and improving the efficiency of vehicle technology. This thesis focuses on the modeling of displacement behavior, based on agent-oriented modeling. We analyzed the data obtained from a survey we conducted for Beirut. We established a relation between employment, social status and mode of transportation selection. The subject of the thesis is an agent-oriented simulation model of road traffic using an institutional economic approach. We propose an architecture, algorithms and methods to implement the agents model in order to model and reproduce their modes of transport behavior. Associated scientific questions are the effects of economic, social, career and household factors on choosing a mode of transport. This work is part of a project of the company S&A (Belgium and Lebanon) to provide tools to enable the transition of means of transport to a more respectful practice.

Les systèmes d'aide à la décision (SAD) doivent permettre aux utilisateurs (décideurs) de prendre les meilleures décisions dans les meilleurs délais. Dans cette thèse, nous nous sommes intéressés aux systèmes qui reposent sur une architecture multi-agents. En effet, les systèmes multi-agents (SMA) permettent de construire des systèmes informatiques ayant recours à l'interrogation multi-critères, souvent utilisée dans les SAD. De façon plus générale, les SMA permettent de concevoir des systèmes qui sont de nature complexe. Cependant, ils n'intègrent pas la notion de contraintes temporelles qui sont souvent très fortes dans les SAD. De plus, dans ces systèmes, des résultats même partiels ou incomplets obtenus dans les temps sont souvent préférés car plus utiles pour la prise de décision que des résultats complets et précis obtenus en retard. Pour cela, les techniques ``anytime'' (raisonnement progressif) semblent une excellente solution. Dans cette thèse, nous présentons une méthode de conception d'un système multi-agent temps réel basé sur l'exploitation des techniques ``anytime''. De plus, nous prenons en compte dans notre modèle l'aspect souvent distribué des SAD.

Dans cette thèse, nous présentons plusieurs contributions pour le routage dans les réseaux tolérants aux délais (DTN). D'abord, nous démontrons, à l'aide d'une analyse de traces réelles, la présence d'hétérogénéité dans les interactions entre les noeuds et que celle-ci peut être prise en compte pour proposer le routage. Deuxièmement, nous proposons l'utilisation d'un formalisme générique basé sur un espace virtuel euclidien, appelé MobySpace, construit à partir d'informations sur les habitudes de mobilité des noeuds. Nous démontrons, avec le rejeux de traces de mobilité réelles, que ce formalisme peut s'appliquer au routage DTN et qu'il permet de créer des stratégies performantes en terme de taux de livraison et de coût de communication. Enfin, nous étudions une architecture de distribution de contenu en environnement urbain à l'aide de bornes courte portée Bluetooth. Nous étudions plusieurs stratégies de distribution en rejouant des traces collectées lors d'une expérience inédite.

Dans le cadre de l'informatique ubiquiste, l'environnement d'exécution d'une application est constitué de machines hétérogènes en ressources matérielles et appartenant à des utilisateurs différents (PC, PDA, téléphone mobile, etc.). Ces caractéristiques imposent de structurer l'application en une organisation d'unités logicielles relativement indépendantes qui coopèrent et interagissent. Dans cette thèse, nous proposons MaDcAr- Agent, un modèle d'agents auto-adaptables à base de composants et muni d'une infrastructure dédiée à l'adaptation. Ce modèle se caractérise par la présence d'un niveau méta qui comporte notamment un moteur d'assemblage en charge des adaptations dynamiques et automatiques en fonction du contexte de l'agent. Le fonctionnement du niveau méta est guidé par la spécification de deux politiques : la politique d'assemblage qui permet à l'agent de s'adapter aux changements de contexte en fonction des composants disponibles et la politique de gestion de contenu qui permet à l'agent d'avoir les composants dont il a le plus besoin grâce aux interactions avec les autres agents. A travers ces spécifications explicites et découplées du comportement applicatif de l'agent, le concepteur d'agents peut prendre en charge la perturbation d'un système dû à des changements imprévus et répétés, sans pour autant nuire à l'autonomie des agents qui composent ce système. Pour valider notre approche, diverses expérimentations ont été menées avec ce modèle, notamment dans le cadre d'un scénario impliquant des robots mobiles qui doivent explorer une zone inconnue.

This thesis focuses on the development of a system in which a team of heterogeneous mobile robots can cooperate to perform a wide range of tasks. In order that a group of heterogeneous robots can cooperate among them, one of the most important parts to develop is the creation of an architecture which gives support for the cooperation. This architecture is developed by means of embedding agents and interfacing agent code with native low-level code. It also addresses the implementation of resource sharing among the whole group of robots, that is, the robots can borrow capabilities from each-other.In order to validate this architecture, some cooperative applications have been implemented. The first one is an application where a group of robots must cooperate in order to safely navigate through an unknown environment. One robot with camera calculates the optical flow values from the images, and from these values calculates the "time to contact" values. This information is shared among the team so that any robot can navigate without colliding with the obstacles.The second cooperative application consists of enabling the team of heterogeneous robots to create a certain formation and navigate maintaining this formation. The application consists of two parts or stages. The first one is the creation of the formation, where a robot with the camera can detect where the rest of the robots are in the environment and indicates to them which is their initial position in the formation. In the second stage the robots must be able to navigate through an environment following the path that the robot with the laser indicates. Due to the odometry errors of the robots, the camera of one of the robots is used so that robots which lose their correct position in the formation can re-align themselves. Finally, in an attempt to facilitate access to the robots of the team and to the information that their accessories provide, a system for the teleoperation of the team has been implemented. This system can be used for teaching robotics or to facilitate the tasks of programming and debugging in the research tasks.

Modeling and simulation of military operations are becoming popular with the widespread application of artificial intelligence methods. As the decision makers would like to analyze the results of the simulations in greater details, entity-level simulation of physical world and activities of actors (soldiers, tanks, etc) is unavoidable. In this thesis, a multiagent framework for simulating task driven autonomous activities of actors or group of actors is proposed. The framework is based on BDI-architecture where an agent is composed of beliefs, goals and plans. Besides, an agent team is organized hierarchically and decisions at different levels of the hierarchy are governed by virtual command agents with their own beliefs, goals and plans. The framework supports an interpreter that realizes execution of single or multiagent plans coherently. The framework is implemented and a case study demonstrating the capabilities of the framework is carried out.