The traffic engineering community currently faces the advent of a new generation of Intelligent Transportation Systems (ITS), known as cooperative systems. More specifically, the recent developments of connected and autonomous vehicles, i.e. cooperative vehicles, are expected to cause a societal shift, changing the way people commute on a daily basis and relate to transport in general. The research presented in this dissertation is motivated by the need for proper understanding of the possible inputs of cooperative vehicles in a traffic stream. Beyond legal aspects regarding the introduction of such vehicles and considerations on standardization and harmonization of the communication norms, the research focuses on the use of communication for highway traffic flow homogenization. In particular, the selected approach for the introduction of cooperation inherits from the theory of traffic flow and the recent developments of microscopic traffic models. Cooperation can first be introduced as a form of multi-anticipation, which can either come from drivers' behaviors or from communication. A mathematical framework for investigating the impact of perturbations into a steady-state traffic is proposed for the class of time continuous car-following models. Linear stability analyses are refined for forward and backward multi-anticipation, exploring the underlying importance of considering upstream information. The linear stability analyses for all wavelengths can be deepened by the mean of the graphical root locus analysis, which enables comparisons and design of strategies of cooperation. The positive influence of bilateral cooperation and of added linear control terms are highlighted. Weakly non-linear analyses are also performed, and the equations of solitary waves appearing at the frontier of the instability domain are obtained. A simple condition over the partial derivatives of the dynamical system is found to determine the acceleration regime of the leading edge of the travelling wave. Following these analytical results, one aim is to simulate a realistic traffic thereby reproducing the driving behavior variability. A Next Generation Simulation trajectory dataset is used to calibrate three continuous car-following models. A methodology involving data filtering, robust calibration, parameters estimation and sampling of realistic parameters is detailed, and allows realistic traffic with stop-and-go waves appearances to be replicated. Based on these simulated trajectories, previous analytical results are confirmed, and the growing perturbations are removed for various coverage rates of cooperative vehicles and adequately tuned cooperative strategies. Finally the issue of information reliability is assessed for a mixed fleet of cooperative and non-cooperative vehicles. The modeling choice consists in building a three layers multi-agent framework that enables the following properties to be defined: the physical behavior of vehicles, the communication possibilities, and the trust each vehicle -or agent- has in another vehicle information or in itself. The investigation of trust and communication rules allow the model to deal with high rates of disturbed cooperative vehicles sensors and to learn in real time the quality of the sent and received information. It is demonstrated that appropriate communication and trust rules sensibly increase the robustness of the network to perturbations coming from exchanges of unreliable information.

Multi-vehicle systems offer many advantages in engineering applications such as increased efficiency and robustness. However, the disadvantage of multi-vehicle systems is that they require a high level of organization and coordination in order to successfully complete a task. Formation control is a field of engineering that addresses this issue, and provides coordination schemes to successfully implement multi-vehicle systems. Two approaches to group coordination were proposed in this work: geometric and self-organizing formations. A geometric reconfiguring formation was developed using the leader-follower method, and the self-organizing formation was developed using the velocity potential equations from fluid flow theory. Both formation controllers were first tested in simulation in MATLAB, and then implemented on the X80 mobile robot units. Various experiments were conducted to test the formations under difficult obstacle scenarios. The robots successfully navigated through the obstacles as a coordinated as a team using the self-organizing and geometric formation control approaches.

Many real-world decision processes require solving optimization problems which may involve different types of constraints such as inequality and equality constraints. The hurdles in solving these Constrained Optimization Problems (COPs) arise from the challenge of searching a huge variable space in order to locate feasible points with acceptable solution quality. Over the last decades Evolutionary Algorithms (EAs) have brought a tremendous advancement in the area of computer science and optimization with their ability to solve various problems. However, EAs have inherent difficulty in dealing with constraints when solving COPs. This thesis presents a new Agent-based Memetic Algorithm (AMA) for solving COPs, where the agents have the ability to independently select a suitable Life Span Learning Process (LSLP) from a set of LSLPs. Each agent represents a candidate solution of the optimization problem and tries to improve its solution through cooperation with other agents. Evolutionary operators consist of only crossover and one of the self-adaptively selected LSLPs. The performance of the proposed algorithm is tested on benchmark problems, and the experimental results show convincing performance. The quality of individuals in the initial population influences the performance of evolutionary algorithms, especially when the feasible region of the constrained optimization problems is very tiny in comparison to the entire search space. This thesis proposes a method that improves the quality of randomly generated initial solutions by sacrificing very little in diversity of the population. The proposed Search Space Reduction Technique (SSRT) is tested using five different existing EAs, including AMA, by solving a number of state-of-the-art test problems and a real world case problem. The experimental results show SSRT improves the solution quality, and speeds up the performance of the algorithms. The handling of equality constraints has long been a difficult issue for evolutionary optimization methods, although several methods are available in the literature for handling functional constraints. In any optimization problems with equality constraints, to satisfy the condition of feasibility and optimality the solution points must lie on each and every equality constraint. This reduces the size of the feasible space and makes it difficult for EAs to locate feasible and optimal solutions. A new Equality Constraint Handling Technique (ECHT) is presented in this thesis, to enhance the performance of AMA in solving constrained optimization problems with equality constraints. The basic concept is to reach a point on the equality constraint from its current position by the selected individual solution and then explore on the constraint landscape. The technique is used as an agent learning process in AMA. The experimental results confirm the improved performance of the proposed algorithm. This thesis also proposes a Modified Genetic Algorithm (MGA) for solving COPs with equality constraints. After achieving inspiring performance in AMA when dealing with equality constraints, the new technique is used in the design of MGA. The experimental results show that the proposed algorithm overcomes the limitations of GA in solving COPs with equality constraints, and provides good quality solutions.

Nous abordons dans cette thèse le problème de la conception d’un système informatique de soutien évaluatif, pour soutenir et évaluer en temps réel l’activité collaborative dans le cas particulier "d'une activité régie par des process issus d'organisations différentes". Nous définissons un process comme un ensemble de règles, de politiques, de plans, de normes ayant pour vocation de guider et de servir de référence à la réalisation d’une activité collaborative. Nous positionnons notre recherche dans le cadre applicatif de l’entrainement à la gestion de crise, et dans le cadre technologique de l’interaction tangible.Pour mettre en place un soutien évaluatif flexible et intelligible, nous proposons un système socio-technique capable d’apporter aux acteurs une conscience organisationnelle partagée et distribuée.Dans ce cadre, trois verrous essentiels sont considérés : (1) représentation et gestion de contextes liés à l’activité soutenue et aux interactions entre acteurs, (2) représentation et gestion des différents process liés à l’activité soutenue et à l’interaction, et (3) articulation entre l'activité soutenue et l'interaction.Pour répondre à ces verrous, nous proposons le système OrA s’appuyant sur trois groupes de principes relatif à l'activité soutenue et à l’interaction : modularité et représentation des process et contextes, gestion autonome et faiblement couplée des process et des éléments de contexte, coordination flexible entre ces mécanismes de gestion de process et de contextes.Ces principes sont mis en oeuvre au sein d’un système informatique s’appuyant sur les modèles des collecticiels, et en particulier CLOVER, des modèles de traces, des systèmes multi-agents normatifs.Ce système est évalué en démontrant la possibilité de modéliser les process d’un cas pratique d’exercice de gestion de crise, puis en proposant une utilisation en situation d’un exercice réel d’entrainement à la gestion de crise. / We discuss in this thesis about designing an computer system for an evaluative support , to support and evaluate in real-time the collaborative activity in the particular case of "an activity governed by processes from different organizations". We define a process as a set of rules, policies, plans, standards which aim to guide and be a reference for the realization of a collaborative activity. Our research is applied to training in crisis management, and situated in the technological context of tangible interaction.To implement a flexible and comprehensible evaluative support, we propose a socio-technical systemfor bringing the actors shared and distributed organizational consciousness.In this context, three key issues are considered: (1) representation and management of the contexts associated with the sustained activity and the interactions between the involved stakeholders, (2) representation and management of various processes associated with the sustained activity and the interaction and (3) the articulation between the sustained activity and the interaction.To answer these issues, we offer the OrA system based on three groups of principles, both about sutained activiy and interaction : modularity and representation of processes and contexts, autonomous management and loosely coupled processes and contextual elements, flexible coordination between these process and context management mechanisms.These principles are implemented in a computer system based on the model of groupware, especially CLOVER, a traces models, and normative multi-agent systems.This system is evaluated by demonstrating the ability to model the process of a practical case of crisis management exercise, and by providing a use situation of a real exercise training in crisis management.

Les études théoriques en épidémiologie utilisent principalement des équations différentielles pour étudier (voire tenter de prévoir) les processus infectieux liés aux maladies contagieuses, souvent sous des hypothèses peu réalistes (ex: des populations spatialement homogènes). Cependant ces modèles ne sont pas bien adaptés pour étudier les processus épidémiologiques à différentes échelles et ils ne sont pas efficaces pour prédire correctement les épidémies. De tels modèles devraient notamment être liés à la structure sociale et spatiale des populations. Dans cette thèse, nous proposons un ensemble de nouveaux modèles dans lesquels différents niveaux de spatialité (par exemple la structure locale de la population, en particulier la dynamique de groupe, la distribution spatiale des individus dans l'environnement, le rôle des personnes résistantes, etc.) sont pris en compte pour expliquer et prédire la façon dont des maladies transmissibles se développent et se répandent à différentes échelles, même à l'échelle de grandes populations. La manière dont les modèles que nous avons développé sont paramétrés leur permet en outre d'être reliés entre eux pour bien décrire en même temps le processus épidémiologique à grande échelle (population d'une grande ville, pays ...) mais avec précision dans des zones de surface limitée (immeubles de bureaux, des écoles). Nous sommes d'abord parvenus à inclure la notion de groupes dans des systèmes d'équations différentielles de modèles SIR (susceptibles, infectés, résistants) par une réécriture des dynamiques de population s'inspirant des réactions enzymatiques avec inhibition non compétitive : les groupes (une forme de complexe) se forment avec des compositions différentes en individus S, I et R, et les individus R se comportent ici comme des inhibiteurs non compétitifs. Nous avons ensuite couplé de tels modèles SIR avec la dynamique globale des groupes simulée par des algorithmes stochastiques dans un espace homogène, ou avec les dynamiques de groupe émergentes obtenues dans des systèmes multi-agents. Comme nos modèles fournissent de l'information bien détaillée à différentes échelles (c'est-à-dire une résolution microscopique en temps, en espace et en population), nous pouvons proposer une analyse de criticité des processus épidémiologiques. Nous pensons en effet que les maladies dans un environnement social et spatial donné présentent des signatures caractéristiques et que de telles mesures pourraient permettre l'identification des facteurs qui modifient leur dynamique.Nous visons ainsi à extraire l'essence des systèmes épidémiologiques réels en utilisant différents modèles mathématique et numériques. Comme nos modèles peuvent prendre en compte les comportements individuels et les dynamiques de population, ils sont en mesure d'utiliser des informations provenant du BigData, collectée par les technologies des réseaux mobiles et sociaux. Un objectif à long terme de ce travail est d'utiliser de tels modèles comme de nouveaux outils pour réduire les épidémies en guidant les rythmes et organisation humaines, par exemple en proposant de nouvelles architectures et en changeant les comportements pour limiter les propagations épidémiques. / Theoretical studies in epidemiology mainly use differential equations, often under unrealistic assumptions (e.g. spatially homogeneous populations), to study the development and spreading of contagious diseases. Such models are not, however, well adapted understanding epidemiological processes at different scales, nor are they efficient for correctly predicting epidemics. Yet, such models should be closely related to the social and spatial structure of populations. In the present thesis, we propose a series of new models in which different levels of spatiality (e.g. local structure of population, in particular group dynamics, spatial distribution of individuals in the environment, role of resistant people, etc) are taken into account, to explain and predict how communicable diseases develop and spread at different scales, even at the scale of large populations. Furthermore, the manner in which our models are parametrised allow them to be connected together so as to describe the epidemiological process at a large scale (population of a big town, country ...) and with accuracy in limited areas (office buildings, schools) at the same time.We first succeed in including the notion of groups in SIR (Susceptible, Infected, Recovered) differential equation systems by a rewriting of the SIR dynamics in the form of an enzymatic reaction in which group-complexes of different composition in S, I and R individuals form and where R people behave as non-competitive inhibitors. Then, global group dynamics simulated by stochastic algorithms in a homogeneous space, as well emerging ones obtained in multi-agent systems, are coupled to such SIR epidemic models. As our group-based models provide fine-grain information (i.e. microscopical resolution of time, space and population) we propose an analysis of criticality of epidemiological processes. We think that diseases in a given social and spatial environment present characteristic signatures and that such measurements could allow the identification of the factors that modify their dynamics.We aim here to extract the essence of real epidemiological systems by using various methods based on different computer-oriented approaches. As our models can take into account individual behaviours and group dynamics, they are able to use big-data information yielded from smart-phone technologies and social networks. As a long term objective derived from the present work, one can expect good predictions in the development of epidemics, but also a tool to reduce epidemics by guiding new environmental architectures and by changing human health-related behaviours.

L'observation des systèmes complexes artificiels à grande échelle est difficile en raison de leur dimension et de leur ouverture. Cette thèse propose le modèle SMA-R (Système Multi-Agent Récursif) basé sur la récursivité pour l'observation multi-niveau des systèmes complexes artificiels. À partir d'un Système Multi-Agent donné, ce modèle est capable de construire une représentation abstraite multi-niveau. Le modèle d'agent récursif proposé possède un module contenant les connaissances, les capacités et le contexte de récursivité; un module pour observer les changements; deux mécanismes pour construire ou détruire les niveaux abstraits; un module d'interaction récursive prenant en charge des comportements collectifs et individuels. Pour appliquer ce modèle à des SMA physiquement décentralisés, nous avons proposé une architecture générique décentralisée d'agents récursifs en adoptant les concepts du modèle OSI qui offre en avant les capacités que nous cherchons: multi niveau, encapsulation, échange de messages virtuels et physiques... Un framework décentralisé générique a été développé permettant aux agents réels de construire les niveaux d'observation. Ce framework est appliqué à l'observation d'un réseau de capteurs sans fil. / Observation of large scale artificials systems is difficult because of their dimension and their openness. This thesis proposes a model SMA-R (Recursive Multi-Agent Systems) based on recursion for multi-level observation of artificial complex systems. From a given SMA, this model is able to build multi-level of abstractions. The model's agent has a module containning knowledge, skills and context of recursion, an eye to observe changes; two mechanisms to build or destroy an abstract level, a module of recursive interaction to support collective and individual behaviors. For applying this model to SMA physically decentralized, we proposed a generic decentralized architecture for recursive agent by adopting the concepts of the OSI model which offerts forward capabilities that we look for: multi-level encapsulation, exchange of virtual and physical messages ... A generic decentralized framework was developed allowing applicatives agents to build multi-levels observation. This framework is applied to the observation of a wireless sensor network.

[pt] A partir do avanço tecnológico, o acesso à informação vem se tornando cada dia mais simples e mais rápido, facilitando consideravelmente o processo de tomada de decisões. Tais mudanças afetam o comportamento de empresas de todas as áreas e da sociedade em geral. No cenário econômico, o Mercado de Valores é um bom exemplo dessa transformação e, por esse motivo, foi escolhido como o domínio de aplicação a ser utilizado pelo simulador Multi-Agent System dor Stock Exchange Simulation (MASSES). Inspirado nos casos de sucesso de competições baseadas em Sistemas Multi-Agentes, trata-se de um simulador onde agentes de software desempenham o papel de investidores. Através do MASSES, podem ser realizados estudos comparativos entre diversas estratégias e análises de como elas se comportam em diferentes situações. A principal contribuição do MASSES é estimular pesquisadores a desenvolver as tecnologias de Engenharia de Software para Sistemas de Multi-Agentes (ESSMA) e inteligência artificial, estreitando ainda mais o relacionamento entre a tecnologia da informação e o mercado financeiro. Essa dissertação apresenta o simulador MASSES e os testes realizados a partir de estratégias de investimento utilizadas no mundo real. / [en] Due to technological advancements in IT, access to information is becoming simpler and faster every day, thus facilitating the decision-making processes. These changes affect the behavior of all kinds of businesses and the society in general. In the economic scenario, Stock Exchange Market is a good example of this transformation and therefore was chosen as the application domain to be explored by a Multi-Agent System for Stock Exchange Simulation (MASSES). Inspired by success stories about Multi-Agent System applied to competitions. MASSES is a simulator where software agents play the role of investors. Through MASSES simulations, it is possible to perform studies among different situations. MASSES main contribution is to encourage researches to develop Software Engineering for Multi-Agents Systems using artificial intelligence techniques, thus strenghtening the relationship between information technology and the financial market. The present dissertation aims to explain MASSES, in addition to showing test results based on the investment strategies used in the real world.

[pt] Sistemas Multiagentes Orientados a Serviços (SOMS) têm surgido visando incorporar os benefícios de duas disciplinas da Engenharia de Software: Arquitetura Orientada a Serviços e Engenharia de Software Orientada a Agentes. A primeira visa fornecer serviços fracamente acoplados que podem ser utilizados em diferentes domínios. A segunda se baseia em um novo paradigma que visa o desenvolvimento de Sistemas Multiagentes, que são compostos por entidades, chamadas agentes, com raciocínio, autonomia e pró-atividade. Um dos principais objetivos de SOMS é ajudar no desenvolvimento de sistemas orientados a serviços capazes de adaptar-se em ambientes computacionais dinâmicos, onde é necessário reagir às mudanças em seus requisitos em tempo de execução, assim como, adaptar-se eficientemente diante de erros de execução e baixa qualidade de serviço. Neste contexto, este trabalho propõe um framework (Java self-Adaptive Agent Framework - JAAF) para implementar agentes autoadaptativos capazes de autonomamente e pró-ativamente descobrir serviços, selecionar o mais apropriado e adaptar-se caso algum problema ocorra durante a execução do serviço. A aplicabilidade do framework proposto é demonstrada através de dois estudos de caso. O primeiro é um sistema responsável por gerar mapas de susceptibilidades, ou seja, mapas que mostram locais com risco de deslizamento em determinada área. O segundo é um sistema onde o principal objetivo é satisfazer necessidades, relacionadas à viagens, de usuários. / [en] Service Oriented Multi-agent Systems (SOMS) have emerged in order to incorporate the benefits of two software engineering disciplines: Serviceoriented Architecture and Agent-oriented Software Engineering. The first provides loosely coupled services that can be used within different domains. The second is based on a new software engineering paradigm that addresses the development of Multi-agent Systems, which are composed of autonomous, pro-active and reasoning entities, named software agents. One of the main goal of SOMS is to help the development of service-oriented systems able to adapt themselves on dynamic computing environments. Those systems must be able to react at runtime to changes in their requirements, as well as to efficiently accommodate for deviations from their expected functionality or quality of services. In this context, this work proposes a framework (Java self- Adaptive Agent Framework - JAAF) to implement self-adaptive agents able to autonomously and pro-actively discover services, decide about the most appropriate one and adapt themselves if they face a problem while executing the service. The applicability of the proposed framework is demonstrated through two case studies. The first is a system responsible for generating susceptibility maps, i.e., maps that show locations with landslides risks in a given area. The second is a system where the main goal is to satisfy the users’ needs related to travel.

O rápido crescimento das regiões urbanas tem impacto significativo nos sistemas de tráfego e transportes. Políticas de gerenciamento e estratégias de planejamento alternativas são claramente necessárias para o tratamento da capacidade limitada, e cada vez mais deficitária, das redes viárias. O conceito de Sistemas Inteligentes de Transportes (ITS) surge neste cenário; mais do que procurar aumentar a capacidade por meio de modificações físicas na infraestrutura, sua premissa baseia-se na utilização de tecnologias avançadas de comunicação e computação para melhor gerir os recursos de tráfego e transportes atuais. Influenciar o padrão do comportamento dos usuários é um desafio que tem estimulado muita pesquisa na área de ITS, onde fatores humanos passam a ter grande importância na modelagem, simulação e avaliação dessa abordagem inovadora. Este trabalho tem como foco a utilização de Sistemas Multiagentes (MAS) na representação dos sistemas de tráfego e transporte, com base nas novasmedidas de desempenho impostas pelas tecnologias ITS. As características de agentes têm grande potencial para representar componentes geográfica e funcionalmente distribuídos, como a maioria dos elementos no domínio da aplicação. Uma arquitetura BDI (beliefs, desires, intentions) é apresentada como alternativa a modelos tradicionais, usados para representar o comportamento do motorista em simulação microscópica, considerando-se a representação explícita dos estados mentais dos usuários. Os conceitos básicos de ITS e MAS são apresentados, assim como exemplos de aplicações relacionados com o tema do trabalho. Esta foi a motivação para a extensão de um simulador microscópico existente, no sentido de incorporar as características dos MAS para melhorar a representação dos motoristas. Assim, a demanda é gerada a partir de uma população de agentes, resultando da decisão sobre a rota e o tempo de partida ao longo de vários dias. O modelo estendido, que passa a suportar a interação de motoristas BDI, foi efetivamente implementado e foram executados diferentes experimentos para testar a abordagem em cenários de tráfego urbano. MAS permite uma abordagem direcionada a processos que facilita a construção de representações modulares, robustas, e extensíveis, características pouco presentes em abordagens voltadas ao resultado. Suas premissas de abstração permitem uma associação direta entre modelo e implementação. Incerteza e variabilidade são assim tratadas de maneira mais intuitiva, uma vez que arquiteturas cognitivas permitem uma fácil representação do comportamento humano na estrutura do motorista. Desta forma, MAS estende a simulação microscópica de tráfego no sentido de melhor representar a complexidade inerente às tecnologias ITS. / The rapid growth of urban areas has a significant impact on traffic and transportation systems. New management policies and planning strategies are clearly necessary to cope with the more than ever limited capacity of existing road networks. The concept of Intelligent Transportation System (ITS) arises in this scenario; rather than attempting to increase road capacity by means of physical modifications to the infrastructure, the premise of ITS relies on the use of advanced communication and computer technologies to handle today’s traffic and transportation facilities. Influencing users’ behaviour patterns is a challenge that has stimulated much research in the ITS field, where human factors start gaining great importance to modelling, simulating, and assessing such an innovative approach. This work is aimed at using Multi-agent Systems (MAS) to represent the traffic and transportation systems in the light of the new performance measures brought about by ITS technologies. Agent features have good potentialities to represent those components of a system that are geographically and functionally distributed, such as most components in traffic and transportation. A BDI (beliefs, desires, and intentions) architecture is presented as an alternative to traditional models used to represent the driver behaviour within microscopic simulation allowing for an explicit representation of users’ mental states. Basic concepts of ITS and MAS are presented, as well as some application examples related to the subject. This has motivated the extension of an existing microscopic simulation framework to incorporate MAS features to enhance the representation of drivers. This way demand is generated from a population of agents as the result of their decisions on route and departure time, on a daily basis. The extended simulation model that now supports the interaction of BDI driver agents was effectively implemented, and different experiments were performed to test this approach in commuter scenarios. MAS provides a process-driven approach that fosters the easy construction of modular, robust, and scalable models, characteristics that lack in former result-driven approaches. Its abstraction premises allow for a closer association between the model and its practical implementation. Uncertainty and variability are addressed in a straightforward manner, as an easier representation of humanlike behaviours within the driver structure is provided by cognitive architectures, such as the BDI approach used in this work. This way MAS extends microscopic simulation of traffic to better address the complexity inherent in ITS technologies.

[pt] Chatbots são programas de computador que interagem com usuários utilizando linguagem natural. Desde sua origem, a tecnologia avançou significantemente e aplicações baseadas na nuvem de grandes empresas permitiram que desenvolvedores criassem chatbots inteligentes e eficientes. No entanto, não há muitas abordagens de desenvolvimento aos principais módulos de um chatbot que são flexíveis o suficiente para permitir a criação de chatbots diferentes para cada domínio, mantendo um robusto controle de diálogo na aplicação. Existem trabalhos que tentam desenvolver uma abordagem mais flexível, cada um com suas vantagens e desvantagens. Uma das vantagens mais notáveis é o uso de sistemas multiagentes para distribuir e realizar tarefas feitas por chatbots. Nesse contexto, este trabalho propõe um framework geral e flexível baseado em sistemas multiagentes para construir chatbots em um domínio escolhido pelo desenvolvedor, com controle de diálogo na aplicação. Esta solução usa uma adaptação da abordagem de estado da informação, e agentes de software, para gestão do diálogo. Para validar a arquitetura proposta, um cenário de uso com 4 chatbots de prova de conceito são analisados e discutidos. / [en] Chatbots are computer programs that interact with users using natural language. Since its inception, the technology has advanced greatly and cloud-based platforms from big companies allow developers to create intelligent and efficient chatbots. However, there are not many development approaches to the main modules of a chatbot that are flexible enough to allow the creation of different chatbots for each domain, while maintaining a robust dialogue control in the application. There have been some works that try to develop a more flexible approach, each of them with their own advantages and disadvantages. One of the most notable advantages is the use of multi-agent systems to distribute and perform the tasks performed by the chatbot. In this context, this work proposes a general and flexible architecture based on multi-agent systems for building chatbots in any domain chosen by the developer, with dialogue control in the application. This architecture uses an adaptation of the information state approach, also using software agents, to perform dialogue management. To validate the proposed architecture, an user scenario involving the implementation of 4 proof of concept chatbots is analyzed discussed.