No description available.

Humans often learn to manipulate objects by observing other people. In much the same way, robots can use imitation learning to pick up useful skills. A system is detailed here for using imitation learning to teach a robot to grasp objects using both hand and whole-body grasps, which use the arms and torso as well as hands. Demonstration grasp trajectories are created by teleoperating a simulated robot to pick up simulated objects. When presented with a new object, the system compares it against the objects in a stored database to pick a demonstrated grasp used on a similar object. Both objects are modeled as a combination of primitives—boxes, cylinders, and spheres—and by considering the new object to be a transformed version of the demonstration object, contact points are mapped from one object to the other. The best kinematically feasible grasp candidate is chosen with the aid of a grasp quality metric. To test the success of the chosen grasp, a full, collision-free grasp trajectory is found and an attempt is made to execute in the simulation. The implemented system successfully picks up 92 out of 100 randomly generated test objects in simulation. / Singapore-MIT Alliance (SMA)

This thesis presents a learning based approach for detecting classes of objects and patterns with variable image appearance but highly predictable image boundaries. It consists of two parts. In part one, we introduce our object and pattern detection approach using a concrete human face detection example. The approach first builds a distribution-based model of the target pattern class in an appropriate feature space to describe the target's variable image appearance. It then learns from examples a similarity measure for matching new patterns against the distribution-based target model. The approach makes few assumptions about the target pattern class and should therefore be fairly general, as long as the target class has predictable image boundaries. Because our object and pattern detection approach is very much learning-based, how well a system eventually performs depends heavily on the quality of training examples it receives. The second part of this thesis looks at how one can select high quality examples for function approximation learning tasks. We propose an {em active learning} formulation for function approximation, and show for three specific approximation function classes, that the active example selection strategy learns its target with fewer data samples than random sampling. We then simplify the original active learning formulation, and show how it leads to a tractable example selection paradigm, suitable for use in many object and pattern detection problems.

This report will deal with the process of creating an example based procedural distribution tool. This is accomplished within the Frostbite game engine editor, FrostEd. By using actual placements of objects in the editor as in-data, the tool provides the artist with an unmatched visual feel for calibrating its properties and settings. Note that this is a unique technique and was invented during the creation of this tool. The tool is based on a machine learning approach. It creates a feature vector from the example placements for each type of object. These vectors are then used to create statistical models which in turn are used to generate new object placements. The process of determining the position and rotation when generating an object is divided into two parts. A new concept called Feature Function (FF) is utilized to provide each element in the population with a probability to obtain a certain position and rotation.

No / Age progression that involves the reconstruction of facial appearance with a natural ageing effect has several applications. These include the search for missing people and identification of fugitives. The majority of age progression methods reported in the literature are data driven. Hence, such methods learn from training data and utilise statistical models such as 3D morphable models and active appearance models (AAM). Principal component analysis (PCA) which is a vital part of these models has an unfortunate drawback of averaging out texture details. Therefore, they work as a low pass filter and as such many of the face skin deformations and minor details become faded. Interestingly, recent work in 2D and 3D animation has shown that patches of the human face are somewhat similar when compared in isolation. Thus, researchers have proposed generating novel faces by compositing small face patches, usually from large image databases. Following these ideas, we propose a novel age progression model which synthesises aged faces using a hybrid of these two techniques. First, an invertible model of age synthesis is developed using AAM and sparse partial least squares regression (sPLS). Then the texture details of the face are enhanced using the patch-based synthesis approach. Our results show that the hybrid algorithm produces both unique and realistic images. Furthermore, our method demonstrates that the identity and ageing effects of subjects can be more emphasised.

In this paper we present a component based person detection system that is capable of detecting frontal, rear and near side views of people, and partially occluded persons in cluttered scenes. The framework that is described here for people is easily applied to other objects as well. The motivation for developing a component based approach is two fold: first, to enhance the performance of person detection systems on frontal and rear views of people and second, to develop a framework that directly addresses the problem of detecting people who are partially occluded or whose body parts blend in with the background. The data classification is handled by several support vector machine classifiers arranged in two layers. This architecture is known as Adaptive Combination of Classifiers (ACC). The system performs very well and is capable of detecting people even when all components of a person are not found. The performance of the system is significantly better than a full body person detector designed along similar lines. This suggests that the improved performance is due to the components based approach and the ACC data classification structure.

L’objectif de cette thèse est la recherche de nouvelles techniques de génération de contenu numérique pour des applications de jeu vidéo. Le manque de variété de terrains, d’objets et de détails affecte fortement le réalisme des paysages de synthèse.Dans ce contexte, un des principaux goulots d’étranglements est la modélisation des ressources graphiques permettant de créer les scènes. Afin de simplifier et d’accélérer cette tâche, nous présentons des méthodes permettant de générer automatiquement du contenu graphique pour créer de grands paysages à la fois complexes et originaux.Notre première approche permet de créer et d’éditer rapidement des variétés d’objets à partir d’un modèle initial fourni par un graphiste, sous la contrainte d’une représentation avec très peu de triangles. Nous présentons également une méthode de génération procédurale des variétés d’objets. Ensemble, ces méthodes permettent de créer aussi bien des variétés de formes naturelles que des ouvrages d’arts tels que des routes, des ponts et des tunnels capables de s’adapter automatiquement au relief d’un paysage.Nous proposons une autre méthode, s’appuyant sur une simulation physique et thermique, pour créer des paysages hivernaux évoluant au cours du temps en fonction des conditions climatiques. Cette approche permet de suivre l’évolution du manteau neigeux ainsi que l’épaisseur de la glace qui se forme en surface d’un lac. / The goal of this thesis is the search for new techniques for generating digital content for video game applications. The lack of variety of objects and details strongly affects the realism of synthetic landscapes. In this context, a major bottleneck is the creation of graphical assets to create the scenes. To simplify and accelerate this task, we present methods to automatically generate graphical content with a view to creating large, convincing and realistic landscapes.We present a first category of method to create and edit quickly a vast variety of objects from an initial model provided by a graphic designer, under the constraint of low polygonal modeling. We also present a procedural method for generating of varieties of objects. Together, these methods can create many varieties of natural forms as well as roads, bridges and tunnels that can automatically adapt to the landform features of an input terrain.We propose another method, based on a physical simulation and heat transfer simulation to create winter landscapes that evolve over time depending on the weather conditions. This approach allows us to follow the evolution of the snowpack and the thickness of the ice that forms on the surface of a lake.

Cette thèse prend pour cadre la synthèse par l'exemple et l'édition de contenu graphique en infographie et propose une réflexion sur les possibles sources d'information utiles à ces fins. Contrairement aux techniques "procédurales", l'approche par l'exemple se distingue par sa grande simplicité d'utilisation : reviennent en effet à l'algorithme de synthèse l'identification, analyse et reproduction des éléments caractéristiques des exemples fournis en entrée par l'utilisateur. Ce mode de création de même que les techniques approfondies d'édition ont grandement contribué à la facilitation de la production à grande échelle de contenus graphiques convaincants et ainsi participé à l'adoption par la communauté des artistes des outils proposés par le support numérique. Mais pour être ainsi exploitées, celles-ci doivent également être hautement contrôlables tout en évitant l'écueil de n'être que le simple prolongement de la main de l'artiste. Nous explorons ici cette thématique dans le cadre de la création de rendus dits expressifs et étudions les interactions (collaboratives ou concurrentielles) entre les différentes sources d'information au cœur de ce processus. Ces dernières sont à notre sens au nombre de trois : l'analyse automatique des données d'entrée avant rendu ou traitement ; l'utilisation de modèles a priori en vue de leur compréhension ; et enfin le contrôle explicite par l'utilisateur. En les combinant au plus juste, nous proposons des techniques nouvelles dans divers domaines de la synthèse en rendu expressif. Au delà du réalisme photographique, le rendu expressif se caractérise par sa poursuite de critères plus difficilement quantifiables tels la facilité de compréhension ou le caractère artistique de ses résultats. La subjectivité de tels objectifs nous force donc ici plus qu'ailleurs à estimer avec soin les sources d'information à privilégier, le niveau d'implication à accorder à l'utilisateur (sans que ce choix ne s'opère au détriment de la qualité théorique de la méthode), ainsi que le possible recours à des modèles d'analyse (sans en compromettre la généralité). Trois principales instances de synthèse sont ici détaillés : la génération de textures, la désaturation d'images, et la représentation de maillages par le dessin au trait. La grande variété des données d'entrée (textures matricielles ou vectorielles, images complexes, géométries 3d), des modalités de synthèse (imitation, conversion, représentation alternative) et d'objectifs (reproduction de la signature visuelle d'une texture, restitution crédible de contrastes chromatiques, génération de dessins conformes au style de l'utilisateur) permettent l'examen de divers équilibres entre ces sources d'information et l'exploration de degrés plus ou moins élevés d'interaction avec l'utilisateur. / This thesis focuses on example-based synthesis and editing in Computer Graphics and reflects on the possible sources of information necessary for such goals. Unlike "procedural" techniques, synthesis by example stands out thanks to its extreme ease-of-use : indeed, tasks such as identification, analysis and reproduction of the distinguishing features of the user-provided examples are left to the method itself. Such approaches, along with today's intricate editing methods have greatly favored the production of compelling graphical content at a wide scale, and henceforth facilitated the adoption of computer-assisted tools by artists. But in order to meet with success, they also have to be highly controllable without being a mere extension of the artist's hand. We explore here such concerns in the context of expressive rendering and study the interactions, may they be collaborative or competitive, between the different sources of information at the core of such processes. In our opinion, there are three main sources of information: the automatic analysis of the inputs before processing; the use of prior knowledge through predetermined models; and users' explicit intervention. Through a clever combination of these sources, we propose new expressive synthesis techniques which satisfy the aforementioned usability. More than photographic realism, expressive rendering strives for the fulfillment of less easily quantifiable goals such as the intelligibility or the aesthetic value of its results. The subjectivity behind the assessment of such criteria thus forces us to attach much importance to the careful choice of the source of information to favor; the required amount of user intervention (without being detrimental to the method's theoretical value); and the possible resort to prior models (without endangering its generality). Three main synthesis instances are studied in this document: texture generation, image de-colorization, and artistic line rendering. The great disparity of inputs (raster and vector textures, complex images, 3d meshes), terms of synthesis (imitation, conversion, depiction) and objectives (preservation of a texture's visual signature, plausible restitution of chromatic contrasts, creation of drawings in accordance with users' styles) gives rise to distinct balances between those sources of information and requires the consideration of various modes of user interaction.

Os artigos científicos são o principal veículo de comunicação para divulgação dos resultados obtidos por milhares de pesquisadores ao redor do mundo. Entretanto, escrever os artigos científicos em publicações de alto impacto é uma tarefa complexa para autores que têm o inglês como segunda língua. Por meio da aprendizagem baseada em exemplos e em experiência, as habilidades de leitura, interpretação e redação de textos científicos podem ser desenvolvidas por intermédio da prática, facilitada por meio de exemplos de trechos de texto reutilizáveis. Para auxiliar na redação de artigos, várias ferramentas computacionais foram desenvolvidas para dar suporte nas várias etapas de concepção do artigo. As ferramentas existentes, que aplicam este tipo de suporte, são todas voltadas para outros temas de pesquisa, incompatíveis com os temas da Engenharia de Produção, que por sua vez possui maneiras particulares de redigir e estruturar o texto. O objetivo da pesquisa desenvolvida consistiu em adaptar e avaliar um instrumento de auxílio à escrita em inglês dirigido a pesquisadores da Engenharia de Produção, fundamentada na aprendizagem baseada em exemplos e em experiências. Para tal fim, foram adaptadas duas ferramentas existentes para auxiliar nas tarefas de leitura, anotação e redação de artigos científicos, avaliadas por meio de dois experimentos com dez alunos de pós-graduação em Engenharia de Produção em universidade pública, nas quais o resultado foi aferido por intermédio de um questionário, buscando identificar os efeitos da utilização sobre o aprendizado e o tempo demandado. O instrumento adaptado oferece, além de exemplos extraídos de artigos publicados, uma estrutura mais coerente com os artigos relacionados à Engenharia de Produção. Os resultados apontam para 90% de aceitação do novo instrumento, em que os alunos relataram um efeito positivo sobre suas habilidades de escrita de artigos, além de uma economia de 11% no tempo dedicado à escrita. Assim sendo, o instrumento tem potencial de auxílio à escrita e sua aplicação traz benefícios reais aos usuários, passíveis de serem adaptadas para qualquer outra área de pesquisa. / Scientific articles are the main communication means of disseminating results obtained by researchers around the world. However, writing scientific articles to highimpact journals is a complex task to authors for whom English is a second language. By means of example- and experience-based learning, reading, interpretation, and writing skills as regards the production of scientific texts can be developed through practice, facilitated by the use of reusable text snippets. Several computational tools have been developed to support the various phases of scientific writing in English. Existing tools that cater to this type of support are all geared to other research topics, incompatible with those inherent to Production Engineering, which has particular features of writing and structuring texts. The purpose of this research was to adapt and assess a new tool to help Production Engineering researchers to write articles English, grounded on example- and experience-based learning. To this end, two existing tools that aid scientists to read, annotate, and write articles in English were adapted and assessed through two experiments with ten Production Engineering graduate students at Brazilian public universities. Data were collected through a questionnaire in order to identify the effect of these tools on participants learning and time required to write articles. The adapted tools provide, in addition to snippets from published articles on Production Engineering, a structure more consistent with this field of knowledge. Results indicate 90% satisfaction rate with the new tool. Also, participants reported a positive effect on their writing skills as well as an 11% reduction in their time spent on scientific writing. Therefore, the tools in question have shown great potential to assist Production Engineering researchers in writing articles; moreover, their features may be adapted to scientific writing in other fields of knowledge.

Os artigos científicos são o principal veículo de comunicação para divulgação dos resultados obtidos por milhares de pesquisadores ao redor do mundo. Entretanto, escrever os artigos científicos em publicações de alto impacto é uma tarefa complexa para autores que têm o inglês como segunda língua. Por meio da aprendizagem baseada em exemplos e em experiência, as habilidades de leitura, interpretação e redação de textos científicos podem ser desenvolvidas por intermédio da prática, facilitada por meio de exemplos de trechos de texto reutilizáveis. Para auxiliar na redação de artigos, várias ferramentas computacionais foram desenvolvidas para dar suporte nas várias etapas de concepção do artigo. As ferramentas existentes, que aplicam este tipo de suporte, são todas voltadas para outros temas de pesquisa, incompatíveis com os temas da Engenharia de Produção, que por sua vez possui maneiras particulares de redigir e estruturar o texto. O objetivo da pesquisa desenvolvida consistiu em adaptar e avaliar um instrumento de auxílio à escrita em inglês dirigido a pesquisadores da Engenharia de Produção, fundamentada na aprendizagem baseada em exemplos e em experiências. Para tal fim, foram adaptadas duas ferramentas existentes para auxiliar nas tarefas de leitura, anotação e redação de artigos científicos, avaliadas por meio de dois experimentos com dez alunos de pós-graduação em Engenharia de Produção em universidade pública, nas quais o resultado foi aferido por intermédio de um questionário, buscando identificar os efeitos da utilização sobre o aprendizado e o tempo demandado. O instrumento adaptado oferece, além de exemplos extraídos de artigos publicados, uma estrutura mais coerente com os artigos relacionados à Engenharia de Produção. Os resultados apontam para 90% de aceitação do novo instrumento, em que os alunos relataram um efeito positivo sobre suas habilidades de escrita de artigos, além de uma economia de 11% no tempo dedicado à escrita. Assim sendo, o instrumento tem potencial de auxílio à escrita e sua aplicação traz benefícios reais aos usuários, passíveis de serem adaptadas para qualquer outra área de pesquisa. / Scientific articles are the main communication means of disseminating results obtained by researchers around the world. However, writing scientific articles to highimpact journals is a complex task to authors for whom English is a second language. By means of example- and experience-based learning, reading, interpretation, and writing skills as regards the production of scientific texts can be developed through practice, facilitated by the use of reusable text snippets. Several computational tools have been developed to support the various phases of scientific writing in English. Existing tools that cater to this type of support are all geared to other research topics, incompatible with those inherent to Production Engineering, which has particular features of writing and structuring texts. The purpose of this research was to adapt and assess a new tool to help Production Engineering researchers to write articles English, grounded on example- and experience-based learning. To this end, two existing tools that aid scientists to read, annotate, and write articles in English were adapted and assessed through two experiments with ten Production Engineering graduate students at Brazilian public universities. Data were collected through a questionnaire in order to identify the effect of these tools on participants learning and time required to write articles. The adapted tools provide, in addition to snippets from published articles on Production Engineering, a structure more consistent with this field of knowledge. Results indicate 90% satisfaction rate with the new tool. Also, participants reported a positive effect on their writing skills as well as an 11% reduction in their time spent on scientific writing. Therefore, the tools in question have shown great potential to assist Production Engineering researchers in writing articles; moreover, their features may be adapted to scientific writing in other fields of knowledge.