Ce travail de recherche est consacré à la formalisation et à la réalisation d'un processus de diagnostic pluridisplinaire. La particularité d'un tel diagnostic résulte du fait qu'il nécessite de nombreux spécialistes, chacun ayant des connaissances sur leur domaine. Le problème principal réside dans les interconnexions entre les domaines. Ces interconnexions peuvent ou non être connues et influer sur le diagnostic. Dans ce manuscrit, nous proposons de réaliser un diagnostic pluridisciplinaire l'aide d'un système multi-agents. Les agents élaborent un diagnostic local à un domaine puis, fusionnent leurs diagnostics afin d'obtenir le diagnostic pluridisciplinaire. Dans ce but, nous proposons un cadre d'argumentation et une méthode de fusion des diagnostics. Ensemble, ces deux propositions forment le modèle ANDi.

Les tuteurs professionnels humains sont capables de prendre en considération des événements du passé et du présent et ont une capacité d'adaptation en fonction d'événements sociaux. Afin d'être considéré comme une technologie valable pour l'amélioration de l'apprentissage humain, un agent cognitif artificiel devrait pouvoir faire de même. Puisque les environnements dynamiques sont en constante évolution, un agent cognitif doit pareillement évoluer et s'adapter aux modifications structurales et aux phénomènes nouveaux. Par conséquent, l'agent cognitif idéal devrait posséder des capacités d'apprentissage similaires à celles que l'on retrouve chez l'être humain ; l'apprentissage émotif, l'apprentissage épisodique, l'apprentissage procédural, et l'apprentissage causal. Cette thèse contribue à l'amélioration des architectures d'agents cognitifs. Elle propose 1) une méthode d'intégration des émotions inspirée du fonctionnement du cerveau; et 2) un ensemble de méthodes d'apprentissage (épisodique, causale, etc.) qui tiennent compte de la dimension émotionnelle. Le modèle proposé que nous avons appelé CELTS (Conscious Emotional Learning Tutoring System) est une extension d'un agent cognitif conscient dans le rôle d'un tutoriel intelligent. Il comporte un module de gestion des émotions qui permet d'attribuer des valences émotionnelles positives ou négatives à chaque événement perçu par l'agent. Deux voies de traitement sont prévues : 1) une voie courte qui permet au système de répondre immédiatement à certains événements sans un traitement approfondis, et 2) une voie longue qui intervient lors de tout événement qui exige la volition. Dans cette perspective, la dimension émotionnelle est considérée dans les processus cognitifs de l'agent pour la prise de décision et l'apprentissage. L'apprentissage épisodique dans CELTS est basé sur la théorie du Multiple Trace Memory consolidation qui postule que lorsque l'on perçoit un événement, l'hippocampe fait une première interprétation et un premier apprentissage. Ensuite, l'information acquise est distribuée aux différents cortex. Selon cette théorie, la reconsolidation de la mémoire dépend toujours de l'hippocampe. Pour simuler de tel processus, nous avons utilisé des techniques de fouille de données qui permettent la recherche de motifs séquentiels fréquents dans les données générées durant chaque cycle cognitif. L'apprentissage causal dans CELTS se produit à l'aide de la mémoire épisodique. Il permet de trouver les causes et les effets possibles entre différents événements. Il est mise en œuvre grâce à des algorithmes de recherche de règles d'associations. Les associations établies sont utilisées pour piloter les interventions tutorielles de CELTS et, par le biais des réponses de l'apprenant, pour évaluer les règles causales découvertes. ______________________________________________________________________________ MOTS-CLÉS DE L’AUTEUR : agents cognitifs, émotions, apprentissage épisodique, apprentissage causal.

Company Purtech in Dals-Ed manufactures molded polyurethane (PUR). Examples of products that include polyurethane are car seats. Robots are used to fill the molds with PUR and to apply the release agent (wax) in the empty molds. Turning from online programming into a graphical offline programming of release agent spraying robots is going to simplify the process by: Applying less of release agent to avoid polluting environment, to produce an easier removal of moulds, for the sake of homogeneous moulds and for economical saving in the cost of release agent Adaption of spraying paths to variation in production speed. Programming of complex spraying trajectories to deal with sharp geometrical subsurface Decreasing onsite programming time (when program a new workpiece or modify an old one); so that robots would be free for production. While turning into offline programming brought the challenges of: Impact of variation in the production speed Lack of 3D models of workcell’s equipments Robot joint configuration when paths and robtargets are in move. Physical Joint limits, Singularities & Reach limits Collisions within the cell space. At the end, the following objectives are successfully met: Adaption of spraying programs to variation in production speed by developing and embedding a method in those programs. Graphical offline generation of spraying trajectories and optimization of those trajectories to the Purtech condition of spraying allowed time for each carrier. Simulation of release agent spraying process; and producing of a well structured RAPID program that reflect the simulated process.

First generation ethanol production has experienced rapid expansion but is now at a crossroads facing impending industry transformation. While Saskatchewan’s ethanol industry has benefited from demand and policy instruments that have guided substantial growth in recent years, changing policy and market dynamics present new challenges which are compelling the industry to adjust. This thesis examines three factors that are suspected to influence ethanol plant locational decisions. The development of an agent-based simulation model in this thesis will ascertain how transportation networks, market synergies, and subsidization influence location stability for an ethanol plant. The long term interaction of these factors is unknown, therefore do tradeoffs exist between these factors or is it conditional for all to be present? Modeling factors that affect location stability through an agent-based approach creates a dynamic framework to understand how location attributes impact an ethanol agent’s longevity. It was found that location stability is affected by an ethanol agent’s distance to both primary transportation networks as well as product markets. Surprisingly, distance to DDGS (dried distillers grain with solubles) markets, a low value by-product of ethanol production, has a profound effect on location stability. Policy instruments and industry subsidization are considered key ethanol development drivers and the surge in ethanol industry growth brought hopes of rural revitalization. In Saskatchewan, policy was developed to support small ethanol plants, those 25 Mmly (million litres per year) or smaller, aimed at increasing farmer investment and alternative markets for wheat. Measuring the effect of subsidization on location stability was fundamental to understanding how a post subsidized ethanol industry may look. The research found that subsidization of Saskatchewan’s ethanol industry dramatically affected economies of scale and location decisions, which left ethanol agents unable to compete in an increasingly competitive ethanol industry.

The technology for the realization of wireless sensors has been available for a long time, but due to progress  and  development  in  electrical  engineering  such  sensors  can  be  manufactured  cost effectively  and  in  large  numbers  nowadays.  This  availability  and  the  possibility  of  creating cooperating  wireless  networks  which  consist  of  such  sensors  nodes,  has  led  to  a  rapidly  growing popularity  of  a  technology  named  Wireless  Sensor  Networks  (WSN).  Its  disadvantage  is  a  high complexity in the task of programming applications based on WSN, which is a result of its distributed and  embedded  characteristic.  To  overcome  this  shortcoming,  software  agents  have  been  identified as  a  suitable  programming  paradigm.  The  agent  based  approach  commonly  uses  a  middleware  for the execution of the software agent. This thesis is meant to compare such agent middleware in their performance in the WSN domain. Therefore two prototypes of applications based on different agent models are implemented for a given set of middleware. After the implementation measurements are extracted  in  various  experiments,  which  give  information  about  the  runtime  performance  of  every middleware in the test set.  In the following analysis it is examined whether each middleware under test  is  suited  for  the  implemented  applications  in  WSN.  Thereupon,  the  results  are  discussed  and compared with the author’s expectations. Finally a short outlook of further possible development and improvements is presented.

We are working on the development and design of an approach to agents that can reason, react to the environment and are able to update their own knowledge as a result of new incoming information. In the resulting framework, rational, reactive agents can dynamically change their own knowledge bases as well as their own goals. An agent can make observations, learn new facts and new rules from the environment, and then update its knowledge accordingly. The knowledge base of an agent and its updating mechanism has been implemented in Logic Programming. The agent’s framework is implemented in Java. This aim of this thesis is to design and implement an architecture of a reactive, rational agent in both Java and Prolog and to test the interaction between the rational part and the reactive part of the agent. The agent architecture is called RR-agent and consists of six more or less components, four implemented in Java and the other two are implemented in XSB Prolog. The result of this thesis is the ground for the paper “An architecture of a rational, reactive agent” by P. DellAcqua, M. Engberg, L.M. Pereira that has been submitted.

The paper introduces an environment for agent-based simulation of corporate strategy. The environment consists of specifications for internal and external factors like cost or market structures that influence corporate strategy, as well as a framework for operational and strategic decisions made by the agents. The entire system is implemented in MATLAB and will be used in a first set of experiments to test the fit of diversification and core competence-based strategies to various types of environment. (author's abstract) / Series: Working Papers SFB "Adaptive Information Systems and Modelling in Economics and Management Science"

This thesis focuses on deadbeat disturbance rejection for discrete-time linear multi-agent systems. The multi-agent systems, on which Spieser and Shams’ decentralized deadbeat output regulation problem is based, are extended by including disturbance agents. Specifically, we assume that there are one or more disturbance agents interacting with the plant agents in some known manner. The disturbance signals are assumed to be unmeasured and, for simplicity, constant. Control agents are introduced to interact with the plant agents, and each control agent is assigned a target plant agent. The goal is to drive the outputs of all plant agents to zero in finite time, despite the presence of the disturbances. In the decentralized deadbeat output regulation problem, two analysis schemes were introduced: targeting analysis, which is used to determine whether or not control laws can be found to regulate, not all the agents, but only the target agents; and growing analysis, which is used to determine the behaviour of all the non-target agents when the control laws are applied. In this thesis these two analyses are adopted to the deadbeat disturbance rejection problem. A new necessary condition for successful disturbance rejection is derived, namely that a control agent must be connected to the same plant agent to which a disturbance agent is connected. This result puts a bound on the minimum number of control agents and constraints the locations of control agents. Then, given the premise that both targeting and growing analyses succeed in the special case where the disturbances are all ignored, a new control approach is proposed for the linear case based on the idea of integral control and the regulation methods of Spieser and Shams. Preliminary studies show that this approach is also suitable for some nonlinear systems.

Tissue, among the highest value added paper products, finds extensive application in modern society. Continued efforts are being made to further improve tissue properties, such as strength, softness and water absorbency. Besides the efforts on characterizing facial tissue softness, this study focuses on tissue quality improvement through chemical means. The application of a wet strength resin, Kymene1500 and a debonding agent, Softrite7516 onto cellulose fibers is considered. First, the adsorption kinetics of the two chemical additives onto cellulose fibers was studied. The adsorption mechanisms were proposed and validated by kinetic data. A novel apparatus was designed in this study, and represented the first in the field to collect real-time data, which has the potential to be applied to the adsorption kinetic study of other types of paper additives. Second, the effects of Kymene1500 and Softrite7516 on various sheet properties were studied. The results provide quantitative information on tissue additives effects on sheet properties. It is shown that the combined application of the additives can overcome the disadvantages of individual species and produce sheets with both wet strength and softness. Finally, environmental-benign debonding agents with polyoxyethylene chains were applied to the sheets, and the effects of two design parameters, i.e., fatty acid and degree of ethoxylation, on tissue properties were investigated.

The main objective of this thesis is to present a new path planning scheme for solving the shortest (collision-free) path problem for an agent (vehicle) operating in a partially known environment. We present two novel algorithms to solve the planning problem. For both of these approaches we assume that the agent has detailed knowledge of the environment and the obstacles only in the vicinity of its current position. Far away obstacles or the final destination are only partially known and may even change dynamically at each instant of time. The path planning scheme is based on information gathered on-line by the available on-board sensor devices. The solution minimizes the total length of the path with respect to a metric that includes actual path length, along with a risk-induced metric. In order to obtain an approximation of the whole configuration space at different levels of fidelity we use a wavelet approximation scheme. In the first proposed algorithm, the path-planning problem is solved using a multi-resolution cell decomposition of the environment obtained from the wavelet transform. In the second algorithm, we extend the results of the the first one by using the multiresolution representation of the environment in conjunction with a conformal mapping to polar coordinates. By performing the cell decomposition in polar coordinates, we can naturally incorporate sector-like cells that are adapted to the data representation collected by the on-board sensor devices.