Un robot autonome collaborant avec des humains doit être capable d'apprendre à se déplacer et à manipuler des objets dans la même tâche. Dans une approche classique, on considère des modules fonctionnels indépendants gérant les différents aspects de la tâche (navigation, contrôle du bras...). A l'opposé, l'objectif de cette thèse est de montrer que l'apprentissage de tâches de natures différentes peut être abordé comme un problème d'apprentissage d'attracteurs sensorimoteurs à partir d'un petit nombre de structures non spécifiques à une tâche donnée. Nous avons donc proposé une architecture qui permet l'apprentissage et l'encodage d'attracteurs pour réaliser aussi bien des tâches de navigation que de contrôle d'un bras.Comme point de départ, nous nous sommes appuyés sur un modèle inspiré des cellules de lieu pour la navigation d'un robot autonome. Des apprentissages en ligne et interactifs de couples lieu/action sont suffisants pour faire émerger des bassins d'attraction permettant à un robot autonome de suivre une trajectoire. En interagissant avec le robot, on peut corriger ou orienter son comportement. Les corrections successives et leur encodage sensorimoteur permettent de définir le bassin d'attraction de la trajectoire. Ma première contribution a été d'étendre ce principe de construction d'attracteurs sensorimoteurs à un contrôle en impédance pour un bras robotique. Lors du maintien d'une posture proprioceptive, les mouvements du bras peuvent être corrigés par une modification en-ligne des commandes motrices exprimées sous la forme d'activations musculaires. Les attracteurs moteurs résultent alors des associations simples entre l'information proprioceptive du bras et ces commandes motrices. Dans un second temps, j'ai montré que le robot pouvait apprendre des attracteursvisuo-moteurs en combinant les informations proprioceptives et visuelles. Le contrôle visuo-moteur correspond à un homéostat qui essaie de maintenir un équilibre entre ces deux informations. Dans le cas d'une information visuelle ambiguë, le robot peut percevoir un stimulus externe (e.g. la main d'un humain) comme étant sa propre pince. Suivant le principe d'homéostasie, le robot agira pour réduire l'incohérence entre cette information externe et son information proprioceptive. Il exhibera alors un comportement d'imitation immédiate des gestes observés. Ce mécanisme d'homéostasie, complété par une mémoire des séquences observées et l'inhibition des actions durant l'observation, permet au robot de réaliser des imitations différées et d'apprendre par observation. Pour des tâches plus complexes, nous avons aussi montré que l'apprentissage de transitions peut servir de support pour l'apprentissage de séquences de gestes, comme c'était le cas pour l'apprentissage de cartes cognitives en navigation. L'utilisation de contextes motivationnels permet alors le choix entre les différentes séquences apprises.Nous avons ensuite abordé le problème de l'intégration dans une même architecture de comportements impliquant une navigation visuomotrice et le contrôle d'un bras robotique pour la préhension d'objets. La difficulté est de pouvoir synchroniser les différentes actions afin que le robot agisse de manière cohérente. Les comportements erronés du robot sont détectés grâce à l'évaluation des actions proposées par le modèle vis à vis des corrections imposées par le professeur humain. Un apprentissage de ces situations sous la forme de contextes multimodaux modulant la sélection d'action permet alors d'adapter le comportement afin que le robot reproduise la tâche désirée.Pour finir, nous présentons les perspectives de ce travail en terme de contrôle sensorimoteur, pour la navigation comme pour le contrôle d'un bras robotique, et son extension aux questions d'interface homme/robot. Nous insistons sur le fait que différents types d'imitation peuvent être le fruit des propriétés émergentes d'une architecture de contrôle sensorimotrice.

La robotique humanoide arrive a maturité avec des robots plus rapides et plus précis. Pour faire face à la complexité mécanique, la recherche a commencé à regarder au-delà du cadre habituel de la robotique, vers les sciences de la vie, afin de mieux organiser le contrôle du mouvement. Cette thése explore le lien entre mouvement humain et le contrôle des systèmes anthropomorphes tels que les robots humanoides. Tout d'abord, en utilisant des méthodes classiques de la robotique, telles que l'optimisation, nous étudions les principes qui sont à la base de mouvements répétitifs humains, tels que ceux effectués lorsqu'on joue au yoyo. Nous nous concentrons ensuite sur la locomotion en nous inspirant de résultats en neurosciences qui mettent en évidence le rôle de la tête dans la marche humaine. En développant une interface permettant à un utilisateur de commander la tête du robot, nous proposons une méthode de contrôle du mouvement corps-complet d'un robot humanoide, incluant la production de pas et permettant au corps de suivre le mouvement de la tête. Cette idée est poursuivie dans l'étude finale dans laquelle nous analysons la locomotion de sujets humains, dirigée vers une cible, afin d'extraire des caractéristiques du mouvement sous forme invariants. En faisant le lien entre la notion "d'invariant" en neurosciences et celle de "tâche cinématique" en robotique humanoide, nous développons une méthode pour produire une locomotion réaliste pour d'autres systèmes anthropomorphes. Dans ce cas, les résultats sont illustrés sur le robot humanoide HRP2 du LAAS-CNRS. La contribution générale de cette thèse est de montrer que, bien que la planification de mouvement pour les robots humanoides peut être traitée par des méthodes classiques de robotique, la production de mouvements réalistes nécessite de combiner ces méthodes à l'observation systématique et formelle du comportement humain.

Building general purpose autonomous robots that suit a wide range of user-specified applications, requires a leap from today's task-specific machines to more flexible and general ones. To achieve this goal, one should move from traditional preprogrammed robots to learning robots that easily can acquire new skills. Learning from Demonstration (LfD) and Imitation Learning (IL), in which the robot learns by observing a human or robot tutor, are among the most popular learning techniques. Showing the robot how to perform a task is often more natural and intuitive than figuring out how to modify a complex control program. However, teaching robots new skills such that they can reproduce the acquired skills under any circumstances, on the right time and in an appropriate way, require good understanding of all challenges in the field. Studies of imitation learning in humans and animals show that several cognitive abilities are engaged to learn new skills correctly. The most remarkable ones are the ability to direct attention to important aspects of demonstrations, and adapting observed actions to the agents own body. Moreover, a clear understanding of the demonstrator's intentions and an ability to generalize to new situations are essential. Once learning is accomplished, various stimuli may trigger the cognitive system to execute new skills that have become part of the robot's repertoire. The goal of this thesis is to develop methods for learning from demonstration that mainly focus on understanding the tutor's intentions, and recognizing which elements of a demonstration need the robot's attention. An architecture containing required cognitive functions for learning and reproduction of high-level aspects of demonstrations is proposed. Several learning methods for directing the robot's attention and identifying relevant information are introduced. The architecture integrates motor actions with concepts, objects and environmental states to ensure correct reproduction of skills. Another major contribution of this thesis is methods to resolve ambiguities in demonstrations where the tutor's intentions are not clearly expressed and several demonstrations are required to infer intentions correctly. The provided solution is inspired by human memory models and priming mechanisms that give the robot clues that increase the probability of inferring intentions correctly. In addition to robot learning, the developed techniques are applied to a shared control system based on visual servoing guided behaviors and priming mechanisms. The architecture and learning methods are applied and evaluated in several real world scenarios that require clear understanding of intentions in the demonstrations. Finally, the developed learning methods are compared, and conditions where each of them has better applicability are discussed. / Att bygga autonoma robotar som passar ett stort antal olika användardefinierade applikationer kräver ett språng från dagens specialiserade maskiner till mer flexibla lösningar. För att nå detta mål, bör man övergå från traditionella förprogrammerade robotar till robotar som själva kan lära sig nya färdigheter. Learning from Demonstration (LfD) och Imitation Learning (IL), där roboten lär sig genom att observera en människa eller en annan robot, är bland de mest populära inlärningsteknikerna. Att visa roboten hur den ska utföra en uppgift är ofta mer naturligt och intuitivt än att modifiera ett komplicerat styrprogram. Men att lära robotar nya färdigheter så att de kan reproducera dem under nya yttre förhållanden, på rätt tid och på ett lämpligt sätt, kräver god förståelse för alla utmaningar inom området. Studier av LfD och IL hos människor och djur visar att flera kognitiva förmågor är inblandade för att lära sig nya färdigheter på rätt sätt. De mest anmärkningsvärda är förmågan att rikta uppmärksamheten på de relevanta aspekterna i en demonstration, och förmågan att anpassa observerade rörelser till robotens egen kropp. Dessutom är det viktigt att ha en klar förståelse av lärarens avsikter, och att ha förmågan att kunna generalisera dem till nya situationer. När en inlärningsfas är slutförd kan stimuli trigga det kognitiva systemet att utföra de nya färdigheter som blivit en del av robotens repertoar. Målet med denna avhandling är att utveckla metoder för LfD som huvudsakligen fokuserar på att förstå lärarens intentioner, och vilka delar av en demonstration som ska ha robotens uppmärksamhet. Den föreslagna arkitekturen innehåller de kognitiva funktioner som behövs för lärande och återgivning av högnivåaspekter av demonstrationer. Flera inlärningsmetoder för att rikta robotens uppmärksamhet och identifiera relevant information föreslås. Arkitekturen integrerar motorkommandon med begrepp, föremål och omgivningens tillstånd för att säkerställa korrekt återgivning av beteenden. Ett annat huvudresultat i denna avhandling rör metoder för att lösa tvetydigheter i demonstrationer, där lärarens intentioner inte är klart uttryckta och flera demonstrationer är nödvändiga för att kunna förutsäga intentioner på ett korrekt sätt. De utvecklade lösningarna är inspirerade av modeller av människors minne, och en primingmekanism används för att ge roboten ledtrådar som kan öka sannolikheten för att intentioner förutsägs på ett korrekt sätt. De utvecklade teknikerna har, i tillägg till robotinlärning, använts i ett halvautomatiskt system (shared control) baserat på visuellt guidade beteenden och primingmekanismer. Arkitekturen och inlärningsteknikerna tillämpas och utvärderas i flera verkliga scenarion som kräver en tydlig förståelse av mänskliga intentioner i demonstrationerna. Slutligen jämförs de utvecklade inlärningsmetoderna, och deras applicerbarhet under olika förhållanden diskuteras. / INTRO

Thèse (D. en sémiologie)--Université du Québec à Montréal, 2005. / En tête du titre: Université du Québec à Montréal. Bibliogr.: f. 276-303. Publié aussi en version électronique.

My project offers a set of comparative close readings of texts by E.T.A. Hoffmann and Honoré de Balzac. Balzac’s early fiction, I contend, grapples with questions relating to the representational practice of mimesis through an explicit engagement with Hoffmann’s work. Hoffmann’s fiction, in turn, proves itself repeatedly to contain the traces of a proto-realist tendency, through its playful interventions into the staging of narrative creation. The contribution of my project to scholarship is twofold. First, it offers comparative readings of texts that have not yet been drawn together, hoping to re-adjust the common ascriptions of ‘Romanticism’ and ‘Realism’ to Hoffmann and Balzac respectively, and to identify a new complication in the relationship of those generic categories. Second, it articulates a new account of mimesis. By drawing on the work of twentieth-century theorists such as Erich Auerbach, Walter Benjamin and Merleau-Ponty, it shows that ‘mimesis’ refers not merely to the imitation of an object, but rather to the reproduction of a particular sensory experience of that object. This perspective on mimesis allows me to unfold new readings of the two authors. How is life compromised in the name of fiction, of the artwork? This question recurs compulsively in Hoffmann’s tales, figured in repeated and near-repeated scenarios in which the everyday is pitted against an ideal or delusional alternative. When Balzac imitates or repeats this mimetic question in the works I consider, it is invariably figured in the image of Hoffmann, called upon as a fictional co-author or authorial double, or as a para-textual element, often in highly visual terms. The thesis thus addresses what I have come to term the ‘stakes of mimesis’. If a particular compromise, or particular stakes, are involved in the creation of fictions, for Balzac those stakes are drawn in distinctly Hoffmannesque terms. The thesis is structured according to the conviction that the relationship between the two writers is not simply a linear one of filiation or influence, but one led by a more complicated sense of imitation. To this end, I take to task the conventional figure of the narrative ‘line’ and follow it through various Romantic and modernist complications. My first chapter, ‘Chiasm’, works as a conceptual introduction to the readings, tracing a particular account of literary mimesis from Plato to Maurice Merleau-Ponty. The four subsequent chapters each read a pair of texts by Hoffmann and Balzac alongside one another. Chapter Two, ‘Line’, focuses on the arabesque lines of Der goldne Topf and La Peau de chagrin. Chapter Three, under the emblem ‘Trope’, examines the paper identities of characters in Die Abenteuer der Sylvester-Nacht and Le Colonel Chabert. Chapter Four, ‘Figure’, considers the delusional artist figures and ekphrastic narrative frameworks of Der Artushof and Le Chef-d’œuvre inconnu. Finally, Chapter Five, ‘Cross’, examines questions of inheritance between Die Elixiere des Teufels and L’Élixir de longue vie. In unfolding these emblematic figures as models of reading, I seek new ways of thinking about the relationship between these two authors, and about the act of comparative reading.

Submitted by Livia Mello (liviacmello@yahoo.com.br) on 2016-09-21T12:29:01Z No. of bitstreams: 1 DissRDBS.pdf: 2067197 bytes, checksum: 5f66defaac763207d5a885501ea67a20 (MD5) / Approved for entry into archive by Marina Freitas (marinapf@ufscar.br) on 2016-09-21T18:33:09Z (GMT) No. of bitstreams: 1 DissRDBS.pdf: 2067197 bytes, checksum: 5f66defaac763207d5a885501ea67a20 (MD5) / Approved for entry into archive by Marina Freitas (marinapf@ufscar.br) on 2016-09-21T18:33:18Z (GMT) No. of bitstreams: 1 DissRDBS.pdf: 2067197 bytes, checksum: 5f66defaac763207d5a885501ea67a20 (MD5) / Made available in DSpace on 2016-09-21T18:33:25Z (GMT). No. of bitstreams: 1 DissRDBS.pdf: 2067197 bytes, checksum: 5f66defaac763207d5a885501ea67a20 (MD5) Previous issue date: 2016-02-17 / Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) / The development and maintenance of vocal verbal repertories that follow structural regularities are processes in need for investigation under the selection by consequences paradigm. The present study investigates the effects of presenting vocal models in passive voice and explicit consequences, of high and medium preference, contingent to descriptions in active voice on the verbal voice of descriptions of Brazilian young children, 4 years old on average. Three experiments were performed. In the first, four children participated. The experimental procedure was composed by five conditions. In the first one, 10 drawings, with two animals interacting, were presented and its descriptions asked. In the second, experimenter and participant alternated in describing 20 drawings, the experimenter always described in passive voice. The third condition was similar to the second, except that descriptions in active voice (divergent) were followed by preferred consequences. The fourth condition was similar to the first one. The fifth condition was similar to the fourth, but it was conducted by an unknown experimenter. All participants described in passive voice after being exposed to the model and continued doing so even when preferred consequences followed descriptions in active voice. But the experimental design may have created a sequence effect between the second and the third conditions. Such limitation was followed in a second experiment. Other four young children participated, 4 years old on average. The procedure was similar to the first experiment, but, the model in passive voice during the second condition was replaced by the presentation of preferred explicit consequences contingent to active voice descriptions. All participants described in passive voice after being exposed to the model, except one. In order to refine the investigation, a third experiment was conducted. Other four young children mean of 4 years old, with different degrees of model following participated. The procedure was identical to the second experiment, but a pre-experimental condition was added in order to measure the participants’ sensibility to the model. Nevertheless, all participants described in passive voice after being exposed to the model in the third experimental condition. The findings point to the importance of investigations on non-arranged contingencies of reinforcement in the development and maintenance of verbal vocal repertoires that follow structural regularities. / O desenvolvimento e manutenção de repertórios verbais vocais que seguem regularidades estruturais são processos pouco investigados sob o paradigma da seleção pelas consequências. O presente estudo investiga os efeitos da apresentação de modelos vocais em voz passiva e de consequências explícitas, de alta e média preferência, contingentes às descrições em voz ativa, sobre a voz verbal das descrições de crianças pequenas, idade média de 4 anos, brasileiras. Três experimentos foram realizados. No primeiro, participaram quatro crianças. O procedimento experimental foi composto por cinco condições. Na primeira condição, 10 desenhos, com dois animais interagindo, foram apresentados e suas descrições solicitadas. Na segunda, experimentador e participante descreveram 20 desenhos alternadamente, sendo as descrições do experimentador sempre em voz passiva. A terceira condição foi semelhante à anterior, porém descrições em voz ativa (divergentes) foram seguidas de consequências preferidas. A quarta condição foi semelhante a primeira. A quinta foi semelhante à quarta, porém foi conduzida por um experimentador desconhecido. Todos os participantes passaram a descrever as figuras em voz passiva após a apresentação do modelo e continuaram a fazê-lo mesmo com apresentação de atividades preferidas contingentes às descrições em voz ativa. Porém, o delineamento empregado pode ter gerado um efeito de sequência entre a segunda e terceira condição. Tal limitação foi investigada no Experimento 2. Participaram outras quatro crianças pequenas, com idades de 4 anos em média. O procedimento foi semelhante ao do primeiro experimento, porém, o modelo em voz passiva durante a segunda condição experimental foi substituído pela apresentação de consequências explícitas preferidas contingentes a descrições em voz ativa. Todos os participantes descreveram os desenhos em voz passiva após serem expostos ao modelo, exceto por um. Buscando refinar a investigação, um terceiro experimento foi realizado. Participaram outras quatro crianças pequenas, média de 4 anos de idade, com diferentes graus de seguimento do modelo. O procedimento foi idêntico ao do segundo experimento, porém, uma tarefa pré-experimental foi adicionada, ela mediu a sensibilidade ao modelo. Não obstante, todos os participantes passaram a falar em voz passiva após serem expostos ao modelo na terceira condição experimental. Os achados apontam para a importância de investigações sobre contingências de reforçamento não arranjadas no desenvolvimento e manutenção de repertórios verbais vocais que seguem regularidades estruturais.

The aim of this thesis is to examine the cognitive processes of social learning from the bottom up. In the field of comparative psychology, an overemphasis on understanding complex cognitive processes in nonhuman animals (e.g. empathy, imitation), may be detrimental to the study of simpler mechanisms. In this thesis, I report five studies of simple cognitive processes related to social learning. A series of experiments with human children and capuchin monkeys (Sapajus sp.), examined action imitation and identified a possible role for associative learning in the development of this ability. An analysis of observational data from captive capuchins explored a number of lesser-studied social learning phenomena, including behavioural synchrony, the neighbour effect, and group-size effects. The results of this study emphasise the importance of exploring behaviour at a number of levels to appreciate the dynamic nature of social influence. Two final experiments examined social contagion in capuchin monkeys, and highlight the importance of describing the relationship between behaviour and emotion to properly understand more complex social cognition. Together, these studies demonstrate how approaching human and nonhuman behaviour from the bottom up, as well as from the top down, can contribute to a better comparative science of social learning.

Robôs sociáveis devem ser capazes de interagir, se comunicar, compreender e se relacionar com os seres humanos de uma maneira natural. Embora diversos robôs sociáveis tenham sido desenvolvidos com sucesso, ainda existem muitas limitações a serem superadas. São necessários importantes avanços no desenvolvimento de mecanismos que possibilitem interações mais realísticas, bem como regulem o relacionamento entre robôs e humanos. Um forma de tornar mais realísticas as interações é através de expressões faciais de emoção. Nesse contexto, este trabalho fornece capacidade de imitação de expressão facial de emoções a uma cabeça robótica virtual, com o objetivo de permitir interações mais realísticas e duradouras com o ser humano. Para isso, é incorporado à mesma aprendizado por imitação, no qual a cabeça robótica imita expressões faciais apresentadas por um usuário durante a interação social. O aprendizado por imitação foi realizado atráves de redes neurais artificiais. As expressões faciais consideradas neste trabalho são: neutra, alegria, raiva, surpresa e tristeza. Os resultados experimentais são apresentados, os quais mostram o bom desempenho do sistema de imitação proposto / Sociable robots must be able to interact, communicate, understand and relate to humans in a natural way. Although many social robots have been developed successfully, there are still many limitations to overcome. Important advances are needed in the development of mechanisms that allow more realistic interactions and that regulate the relationship between robots and humans. One way to make more realistic interactions is through facial expressions of emotion. In this context, this project provides ability for imitation of facial expressions of emotion to a virtual robotic head, in order to allow more realistic and lasting interactions with humans. For such, learning by imitation is used, in which the robotic head mimics facial expressions made by a user during social interaction. The imitation learning was performed by artificial neural networks. Facial expressions considered in this work are: neutral, happiness, anger, surprise and sadness. Experimental results are presented which show the good performance of the proposed system imitation

Em Lucíola, José de Alencar assume uma posição peculiar de autor real, na medida em que contrasta com o Sr. Paulo Silva, autor designado na ficção. Narrador de seus conflitos, o moço espelha um herói confessional, examinando o próprio passado em busca da imagem de uma mulher. Senhora GM é sua interlocutora. Lúcia se circunscreve como modelo, pessoa retratada, namorada e cortesã. Esta análise está mais interessada em como as coisas são ditas e menos no que é dito, de sorte que se dedica à investigação dos mecanismos da enunciação. O estudo não focaliza a personagem feminina, ao contrário está voltado para o amante; e muito embora não seja contra o homem, ambiciona, através do primeiros capítulos, confrontar sua autoimagem e a interpretação que ele oferece para a história. O quinto segmento desta leitura questiona os modos através dos quais palavras e imagens dão forma ao sexo na intimidade do texto. O conjunto seguinte explora a questão da imitação como um tema recorrente no livro e como um fio que costura forma e conteúdo, palavra e ideia. A última seção investiga os nomes das personagens enquanto uma escolha de palavras, um procedimento no domínio do artifício. / In Lucíola, José de Alencar is the real author whereas Mr. Paulo Silva is the fictional one. The narrator is also a confessional hero examining his own past searching for the image of a woman. Mrs. GM is his interlocutor. Lúcia is the sitter, the one that is being portrayed, girlfriend and courtesan. This analysis is more interested in how things are said than in what is said; therefore it contemplates the mechanisms of enunciation. The study is not focused in the female character but dedicated to her lover; even though it is not against the man, it aims, throughout the first chapters, to confront his self-image and the interpretation he offers to the story. The fifth segment interrogates how words and imagery give a form to sex as a matter of speech. The following portion explores imitation as a recurrent theme in the book and as a thread that sews form and meaning, words and ideas. The last section investigates the names of the characters as a choice of words, under the domain of artifice.

Context. Developing an Artificial Intelligence (AI) agent that canpredict and act in all possible situations in the dynamic environmentsthat modern video games often consists of is on beforehand nearly im-possible and would cost a lot of money and time to create by hand. Bycreating a learning AI agent that could learn by itself by studying itsenvironment with the help of Reinforcement Learning (RL) it wouldsimplify this task. Another wanted feature that often is required is AIagents with a natural acting behavior and a try to solve that problemcould be to imitating a human by using Imitation Learning (IL). Objectives. The purpose of this investigation is to study if it is pos-sible to create a learning AI agent feasible to play and complete somelevels in a platform game with the combination of the two learningtechniques RL and IL. Methods. To be able to investigate the research question an imple-mentation is done that combines one RL technique and one IL tech-nique. By letting a set of human players play the game their behavioris saved and applied to the agents. The RL is then used to train andtweak the agents playing performance. A couple of experiments areexecuted to evaluate the differences between the trained agents againsttheir respective human teacher. Results. The results of these experiments showed promising indica-tions that the agents during different phases of the experiments hadsimilarly behavior compared to their human trainers. The agents alsoperformed well when comparing them to other already existing ones. Conclusions. To conclude there is promising results of creating dy-namical agents with natural behavior with the combination of RL andIL and that it with additional adjustments would make it performeven better as a learning AI with a more natural behavior.