Les systèmes de vision artificielle sont aujourd'hui limités à l'extraction de données issues de ce que les caméras « voient ». Cependant, la compréhension de ce qu'elles voient peut être enrichie en associant la connaissance du contexte et la connaissance d'interprétation d'un humain.Dans ces travaux de thèse, nous proposons une approche associant des algorithmes de vision atificielle à une modélisation sémantique du contexte d'acquisition.Cette approche permet de réaliser un raisonnement sur la connaissance extraite des images par les caméras en temps réel. Ce raisonnement offre une réponse aux problèmes d'occlusion et d'erreurs de détections inhérents aux algorithmes de vision artificielle. Le système complet permet d'offrir un ensemble de services intelligents (guidage, comptage...) tout en respectant la vie privée des personnes observées. Ces travaux forment la première étape du développement d'un bâtiment intelligent qui peut automatiquement réagir et évoluer en observant l'activité de ces usagers, i.e., un bâtiment intelligent qui prend en compte les informations contextuelles.Le résultat, nommé WiseNET, est une intelligence artificielle en charge des décisions au niveau du bâtiment (qui pourrait être étendu à un groupe de bâtiments ou même a l'échelle d'un ville intelligente). Elle est aussi capable de dialoguer avec l'utilisateur ou l'administrateur humain de manière explicite. / To date, computer vision systems are limited to extract digital data of what the cameras "see". However, the meaning of what they observe could be greatly enhanced by environment and human-skills knowledge.In this work, we propose a new approach to cross-fertilize computer vision with contextual information, based on semantic modelization defined by an expert.This approach extracts the knowledge from images and uses it to perform real-time reasoning according to the contextual information, events of interest and logic rules. The reasoning with image knowledge allows to overcome some problems of computer vision such as occlusion and missed detections and to offer services such as people guidance and people counting. The proposed approach is the first step to develop an "all-seeing" smart building that can automatically react according to its evolving information, i.e., a context-aware smart building.The proposed framework, named WiseNET, is an artificial intelligence (AI) that is in charge of taking decisions in a smart building (which can be extended to a group of buildings or even a smart city). This AI enables the communication between the building itself and its users to be achieved by using a language understandable by humans.

Submitted by Mariana Dornelles Vargas (marianadv) on 2015-05-25T12:29:15Z No. of bitstreams: 1 AutoEduMat.pdf: 1060362 bytes, checksum: 25b8156de4b9c2c2c5b9dc0f69aea011 (MD5) / Made available in DSpace on 2015-05-25T12:29:15Z (GMT). No. of bitstreams: 1 AutoEduMat.pdf: 1060362 bytes, checksum: 25b8156de4b9c2c2c5b9dc0f69aea011 (MD5) Previous issue date: 2010 / Nenhuma / Esta dissertação apresenta uma pesquisa relacionada as ferramentas que dão suporte a utilização de objetos de aprendizagem em plataformas digitais. Mais especificamente, a pesquisa se direciona para as ferramentas de apoio a autoria destes objetos, em particular dos seus metadados. Inicialmente é apresentada a contextualização do problema de pesquisa, sua fundamentação teórica e os trabalhos relacionados ao tema. Em seguida são apresentadas as principais características do sistema proposto, o AutoEduMat - Ferramenta de Apoio a Autoria de Metadados de Objetos de Aprendizagem para o Domínio de Ensino de Matemática. A ferramenta AutoEduMat dá apoio a autoria de objetos de aprendizagem, oferecendo assistência ao projetista (designer) de objetos na criação e edição de metadados destes objetos. A principal inovação do trabalho é a combinação das tecnologias de Engenharia de Software de Agentes e de Engenharia de Ontologias para construir um sistema multiagente que oferece suporte inteligente para a geração dos metadados dos objetos de aprendizagem, sendo capaz de interagir com o usuário com termos de seu próprio contexto profissional e educacional. No trabalho é proposta a ontologia Onto-EduMat que incorpora os conhecimentos sobre o domínio de ensino de matemática, incluindo aspectos pedagógicos, necessários para o auxílio a geração dos metadados. Tanto a ferramenta quanto seu modelo ontológico são validados através de experimentos descritos no final do trabalho. / This dissertation presents a research related to the tools that support the utilization of learning objects in digital platforms. More precisely, the research is directed to the tools that support the authoring process of these objects, in particular of their metadata. Initially are presented the characterization of the problem, its theoretical foundations and related works. Then are presented the main characteristics of the proposed system, the AutoEduMat - Metadata Authoring Tool for Mathematics Learning Objects. The AutoEduMat system will provide assistance to the object designer in the metadata creation and edition of these objects. The main innovation of this work is the combination of Agent Oriented Software Engineering and Ontology Engineering technologies to built a multiagent system able to offer intelligent support for metadata creation, interacting with users using terms related to their professional and educational context. This work proposes the Onto-EduMat ontology, which incorporates the mathematical and pedagogical knowledge necessary to generate the metadata. The authoring tool and its ontological model are validated through experiments described in the end of the work.

La « conception pour la fabrication » ou DFM (Design for Manufacturing) est une méthode maintenant classique pour assurer lors de la conception des produits simultanément la faisabilité, la qualité et le rendement de la production. Dans l'industrie microélectronique, le Design Rule Manual (DRM) a bien fonctionné jusqu'à la technologie 250nm avec la prise en compte des variations systématiques dans les règles et/ou des modèles basés sur l'analyse des causes profondes, mais au-delà de cette technologie, des limites ont été atteintes en raison de l'incapacité à sasir les corrélations entre variations spatiales. D'autre part, l'évolution rapide des produits et des technologies contraint à une mise à jour « dynamique » des DRM en fonction des améliorations trouvées dans les fabs. Dans ce contexte les contributions de thèse sont (i) une définition interdisciplinaire des AMDEC et analyse de risques pour contribuer aux défis du DFM dynamique, (ii) un modèle MAM (mapping and alignment model) de localisation spatiale pour les données de tests, (iii) un référentiel de données basé sur une ontologie ROMMII (referential ontology Meta model for information integration) pour effectuer le mapping entre des données hétérogènes issues de sources variées et (iv) un modèle SPM (spatial positioning model) qui vise à intégrer les facteurs spatiaux dans les méthodes DFM de la microélectronique, pour effectuer une analyse précise et la modélisation des variations spatiales basées sur l'exploitation dynamique des données de fabrication avec des volumétries importantes. / The DFM (design for manufacturing) methods are used during technology alignment and adoption processes in the semiconductor industry (SI) for manufacturability and yield assessments. These methods have worked well till 250nm technology for the transformation of systematic variations into rules and/or models based on the single-source data analyses, but beyond this technology they have turned into ineffective R&D efforts. The reason for this is our inability to capture newly emerging spatial variations. It has led an exponential increase in technology lead times and costs that must be addressed; hence, objectively in this thesis we are focused on identifying and removing causes associated with the DFM ineffectiveness. The fabless, foundry and traditional integrated device manufacturer (IDM) business models are first analyzed to see coherence against a recent shift in business objectives from time-to-market (T2M) and time-to-volume towards (T2V) towards ramp-up rate. The increasing technology lead times and costs are identified as a big challenge in achieving quick ramp-up rates; hence, an extended IDM (e-IDM) business model is proposed to support quick ramp-up rates which is based on improving the DFM ineffectiveness followed by its smooth integration. We have found (i) single-source analyses and (ii) inability to exploit huge manufacturing data volumes as core limiting factors (failure modes) towards DFM ineffectiveness during technology alignment and adoption efforts within an IDM. The causes for single-source root cause analysis are identified as the (i) varying metrology reference frames and (ii) test structures orientations that require wafer rotation prior to the measurements, resulting in varying metrology coordinates (die/site level mismatches). A generic coordinates mapping and alignment model (MAM) is proposed to remove these die/site level mismatches, however to accurately capture the emerging spatial variations, we have proposed a spatial positioning model (SPM) to perform multi-source parametric correlation based on the shortest distance between respective test structures used to measure the parameters. The (i) unstructured model evolution, (ii) ontology issues and (iii) missing links among production databases are found as causes towards our inability to exploit huge manufacturing data volumes. The ROMMII (referential ontology Meta model for information integration) framework is then proposed to remove these issues and enable the dynamic and efficient multi-source root cause analyses. An interdisciplinary failure mode effect analysis (i-FMEA) methodology is also proposed to find cyclic failure modes and causes across the business functions which require generic solutions rather than operational fixes for improvement. The proposed e-IDM, MAM, SPM, and ROMMII framework results in accurate analysis and modeling of emerging spatial variations based on dynamic exploitation of the huge manufacturing data volumes.