Zugl.: Braunschweig, Techn. University, Diss., 2009.

The elderly population worldwide has an increasing expectation of wellbeing and life expectancy. The monitoring of the majority of elderly people on an individual basis, in a medical sense, will not be a viable proposition in the future due to the projected numbers of individuals requiring such activity. The expectation is that the infrastructure available will not be adequate to meet all the anticipated requirements and subsequently people will have to live at home with inadequate care. A new global objective that aims towards enhancing the quality of life of the elderly is being supported by extensive research. This research has been taking place in the field of ambient intelligence (AmI), considering factors including more comfort, improved health, enhanced security for the elderly, and facilitating the living in their homes longer. Prior research has shown a need for accelerated expansion in the ambient intelligence domain. To that end this work presents a novel learning technique for intelligent agents that can be used in Ambient Intelligent Environments (AIEs). The main objective of this work is to add knowledge to the AmI domain and to explore the practical applications within this research field. The added knowledge is accomplished through the development of an ambient intelligent health care environment that allows a practical assessment of the human well-being to take place. This is achieved by transforming the elderly living environment into an intelligent pseudo robot within which they reside to better understand the human wellbeing. The system developed aims to provide evidence that a level of automated care is both possible and practical. This care is for those with chronic physical or mental disabilities who have difficulty in their interactions with standardised living spaces. The novel integrated hardware and software architecture provides personalised environmental monitoring. It also provides control facilities based on the patient‘s physical and emotional wellness in their home. Entitled Health Adaptive Online Emotion Fuzzy Agent (HAOEFA), the system provides a non-invasive, self-learning, intelligent controlling system that constantly adapts to the requirements of an individual. The system has the ability to model and learn the user behaviour in order to control the environment on their behalf. This is achieved with respect to the changing environmental conditions as well as the user‘s health and emotional states being detected. A change of emotion can have a direct impact on the system‘s control taking place in the environment. Thus HAOEFA combines an emotion recognition system within a fuzzy logic learning and adaptation based controller. The emotion recogniser detects the occupant‘s emotions upon the changes of the physiological data being monitored. In addition to acting as an output to the occupant‘s physiological changes, the detected emotion also acts as input to the whole situation being observed by HAOEFA. This allows HAOEFA to control the Glam i-HomeCare on the user‘s behalf with respect to their emotional status. The system developed incorporates real-time, continuous adaptations to facilitate any changes to the occupant‘s behaviour within the environment. It also allows the rules to be adapted and extended online, assisting a life-long learning technique as the environmental conditions change and the user behaviour adjusts with it. HAOEFA uses the fuzzy c-means clustering methodology for extracting membership functions (MFs) before building its set of fuzzy rules. These MFs together with the rules base constitute a major part of the proposed system. It has the ability to learn and model the individual human behaviour with respect to their emotional status. Following the provided literature review and the presentation of Fuzzy logic MFs (see section 3.3). The thesis presents two chosen unobtrusive self-learning techniques that are used in the development of the intelligent fuzzy system. Each approach combines an emotion recogniser with a fuzzy logic learning and adaptation based technique for systems that can be used in AIEs. A comparison of two different MFs designs is contrasted showing the impact they have on the system learning ability. A number of carefully designed experiments were performed by volunteers in the Glam i-HomeCare test-bed at the University of South Wales to examine the system‘s ability to learn the occupant‘s behaviour with respect to their health and emotional states. The experimental procedures were performed twice by each volunteer, while maintaining the same behavioural actions to compare how much the design of fuzzy membership functions can impact the learning process and the number of rules created by the system. Besides evaluating both systems‘ emotion recognition accuracies and comparing them to one another for each occupant, the empirical outcomes show the potential of the approach in assisting the extension of independent living. The results demonstrate how the type-1 fuzzy system both learnt and adapted to each occupant‘s behaviour with respect to their health and emotional state whilst assessing multiple environmental conditions.

La notion de confort dans les habitats est une problématique majeure pour résoudre des problèmes écologiques (consommation et émissions des bâtiments), économiques (réduction de coûts d'exploitation) et sociaux (maintien et assistance à domicile) qui définissent le développement durable. Cependant, cette notion de confort est complexe, par le nombre de paramètres qu'elle intègre, paramètres à la fois humains (perception) et physiques (mesure). Notre étude vise à modéliser cette notion de confort dans un contexte d'habitat intelligent. L'habitat intelligent émerge depuis le début des années 2000, et se positionne en héritier de la domotique, bénéficiant des progrès technologiques illustrés par l'informatique ubiquitaire et l'intelligence artificielle, concepts formants l'intelligence ambiante. La première partie de notre étude consiste à définir l'habitat intelligent, en formalisant les acquis (domotique) et les problématiques de recherche, sous l'angle de la représentation de connaissances par les modèles. Notre approche du bâtiment intelligent nous à permis de définir un cadre d'interopérabilité : un intergiciel capable de concentrer les paramètres et commandes d'un environnement. Cette interopérabilité est nécessaire de par l'hétérogénéité des objets communicants qui composent un habitat : hétérogénéité des applications, des protocoles de communication, de savoir-faire et d'usages. Les travaux réalisés dans cette première partie de l'étude nous ont permis d'instrumenter une plate-forme d'expérimentation : la plateforme Domus. Ainsi, en reconstituant un appartement, et en le dotant d'objets communicants, nous avons pu mettre en œuvre, par le biais de l'interopérabilité, un environnement intelligent, environnement qui se caractérise par une forte densité d'information et une capacité de réaction. La réalisation de cette plate-forme est nécessaire pour aborder des thématiques diverses liées à l'habitat, comme le confort. En effet, l'intelligence ambiante apporte une nouvelle dimension dans ce cadre de recherche : l'ubiquité. La densité croissante de capteurs nous permet de collecter plus d'informations, non seulement sur l'environnement mais également sur l'utilisateur et son comportement, définissant ainsi une nouvelle approche du confort : le confort adaptatif. Les travaux sur l'étude du confort dans les bâtiments se focalisent sur le confort thermique. Dans nos travaux, nous avons voulu nous intéresser au confort multi-sensoriel. Celui-ci permet d'une part de prendre en compte l'ensemble des paramètres qui agrémentent un environnement (l'air, le son, la vue) mais permet également de nous intéresser aux effets sensoriels croisés que peuvent induire ces modalités sur l'occupant. Par exemple, on soupçonne la température d'éclairage (éclairage rouge/chaud, éclairage bleu/froid) d'avoir une incidence sur la perception thermique. Des expérimentations ont en effet démontré l'approche pratique et l'approche théorique de ces effets multi-sensoriels. La mise en place de notre cadre d'interopérabilité, en première partie, dans la plateforme Domus et les résultats de nos évaluations expérimentales, en seconde partie, sur le confort réalisés dans cette même plateforme, nous permettent de participer à la définition d'un « confort-mètre », qui s'appuie à la fois sur les capteurs, les objets de l'habitat et la perception des habitants. / The notion of comfort in homes is a major problem to solve environmental problems (consumption and emissions of buildings), economic (reduction of operating costs) and social (maintenance and home care) that define sustainable development. However, this notion of comfort is complicated by the number of parameters that integrates both human (perception) and physical (measurement) parameters. Our study aims to model the concept of comfort in a smart home. Smart homes emerged in the early 2000s, and are positioned as heir to home automation, benefiting from technological advances illustrated by ubiquitous computing and artificial intelligence, ambient intelligence concepts formants. The first part of this study was to define habitat intelligent, formalizing the gains (home automation) and research issues, in terms of knowledge representation by the models. Our approach to intelligent building allowed us to define a framework for interoperability: a middleware able to focus and control the parameters of an environment. This interoperability is required by the heterogeneity of communicating objects that make up a habitat: heterogeneity of applications, communication protocols, know-how and practices. This first part of the study allowed us to instrument an experimental platform: the platform Domus. Thus, by restoring an apartment, and by providing it with smart objects, we could implement, through interoperability, an intelligent environment, environment characterized by high information density and capacity reaction. The realization of this platform is needed to address various topics related to housing, such as comfort. Indeed, ambient intelligence brings a new dimension in this research framework: ubiquity. The increasing density of sensors allows us to collect more information, not only the environment but also on the user and its behavior, thus defining a new approach to comfort: Adaptive comfort. Most of the work focusing on thermal comfort, we are interested in multi-sensory comfort. This allows one hand to take into account all the parameters that enhances an environment (air, sound, sight) but also allows our attention to cross-sensory effects that can induce these terms on the occupier. For example, it is suspected the temperature light (red light / heat, light blue / cold) to affect the perception of heat. Experiments have shown the practical approach and the theoretical approach of multi-sensory effects. The results of this study will be led to participate in the definition of "comfortmeter", a tool to sense comfort for both habitat and the inhabitant.

La Maison Intelligente est une résidence équipée de technologie informatique qui assiste ses habitant dans les situations diverses de la vie domestique en essayant de gérer de manière optimale leur confort et leur sécurité par action sur la maison. La détection des situations anormales est un des points essentiels d'un système de surveillance à domicile. Ces situations peuvent être détectées en analysant les primitives générées par les étages de traitement audio et par les capteurs de l'appartement. Par exemple, la détection de cris et de bruits sourds (chute d'un objet lourd) dans un intervalle de temps réduit permet d'inférer l'occurrence d'une chute. Le but des travaux de cette thèse est la réalisation d'un contrôleur intelligent relié à tous les périphériques de la maison capable de réagir aux demandes de l'habitant (par commande vocale) et de reconnaître des situations à risque ou détresse. Pour accomplir cet objectif, il est nécessaire de représenter formellement et raisonner sur des informations, le plus souvent temporelles, à des niveaux d'abstraction différents. Le principale défi est le traitement de l'incertitude, l'imprécision, et incomplétude, qui caractérisent les informations dans ce domaine d'application. Par ailleurs, les décisions prises par le contrôleur doivent tenir compte du contexte dans lequel une ordre est donné, ce qui nous place dans l'informatique sensible au contexte. Le contexte est composé des informations de haut niveau tels que la localisation, l'activité en cours de réalisation, la période de la journée. Les recherches présentées dans ce manuscrit peuvent être divisés principalement en trois axes: la réalisation des méthodes d'inférence pour acquérir les informations du contexte(notamment, la localisation de l'habitant y l'activité en cours) à partir des informations incertains, la représentation des connaissances sur l'environnement et les situations à risque, et finalement la prise de décision à partir des informations contextuelles. La dernière partie du manuscrit expose les résultats de la validation des méthodes proposées par des évaluations amenées à la plateforme expérimental Domus. / A smart home is a residence featuring ambient intelligence technologies in order to help its dwellers in different situations of common life by trying to manage their comfort and security through the execution of actions over the effectors of the house. Detection of abnormal situations is paramount in the development of surveillance systems. These situations can be detected by the analysis of the traces resulting from audio processing and the data provided by the network of sensors installed in the smart home. For instance, detection of cries along with thuds(fall of a heavy object) in a short time interval can help to infer that the resident has fallen. The goal of the research presented in this thesis is the implementation of an intelligence controller connected with the devices in the house that is able to react to user's commands(through vocal interfaces) and recognize dangerous situations. In order to fulfill this goal, it is necessary to create formal representation and to develop reasoning mechanism over informations that are often temporal and having different levels of abstraction. The main challenge is the processing the uncertainty, imprecision, and incompleteness that characterise this domain of application. Moreover, the decisions taken by the intelligent controller must consider the context in which a user command is given, so this work is made in the area of Context Aware Computing. Context includes high level information such as the location of the dweller, the activity she is making, and the time of the day. The research works presented in this thesis can be divided mainly in three parts: the implementation of inference methods to obtain context information(namely, location and activity) from uncertain information, knowledge representation about the environment and dangerous situations, and finally the development of decision making models that use the inferred context information. The last part of this thesis shows the results from the validation of the proposed methods through experiments performed in an experimental platform, the Domus apartment.

Il y a près de 20 ans, Marc Weiser a imaginé l'ordinateur du 21ème siècle et a proposé le concept de l’informatique ubiquitaire. Une des idées principales de Weiser a récemment évolué vers un paradigme plus général connu comme la sensibilité au contexte qui est devenu un thème très important en informatique ubiquitaire. Depuis Active Badge considéré comme la première application sensible au contexte, de nombreuses tentatives pour construire des systèmes sensibles au contexte efficaces ont vu le jour. Cependant la problématique comment acquérir contexte, comment le traiter et comment créer des applications sensibles au contexte est encore aujourd’hui un défi suscitant de nombreuses recherches. Cette thèse étudie en profondeur certaines questions clés liées à la sensibilité au contexte et au développement d’intergiciels sensibles au contexte. Les principales contributions de notre recherche concernent la prise en compte du contexte spatiotemporel et sa modélisation, la conception et l’implémentation d’un intergiciel sensible au contexte et d’un moteur intelligent d'inférence de contexte. Le modèle de représentation du contexte spatiotemporel proposé vise à organiser le contexte et ses relations pour un système sensible au contexte. La méthode basée sur les ontologies est adoptée pour construire notre modèle, supportant à la fois le partage des connaissances et leur réutilisation ainsi que la déduction logique. Ce modèle adopte une structure hiérarchique à deux couches pour modéliser les situations à prendre en compte. La couche supérieure s’occupe du contexte commun générique, tandis que la couche inférieure se concentre sur les caractéristiques plus spécifiques. A la différence des modèles existants, en plus de prendre en compte l’aspect localisation, notre modèle prend également en charge la gestion d’historique des de contextes en s’appuyant sur différentes ressources. Ces historiques de contexte peuvent être utilisés pour prévoir et inférer le contexte. Un middleware sensible au contexte a été conçu comme une plateforme permettant la récupération et le traitement du contexte. Elle est organisée en deux couches : La couche basse apporte une solution à l’intégration des capteurs et actionneurs avec une représentation de données normalisée ; la couche haute propose un interpréteur de contexte polyvalent qui s’appuie sur quatre éléments : un agrégateur de contexte, un moteur d'inférence, une base de connaissance de contextes et un moteur de recherche en charge de la déduction de contexte, de l’interrogation et du stockage persistant. Ce middleware fournit un environnement pour le prototypage rapide de services sensibles au contexte pour l’intelligente ambiante. Le moteur intelligent d’inférence est le composant central de notre middleware. Pour le concevoir nous avons d’abord étudié toutes les méthodes publiées pendant les dix dernières années dans les trois conférences de premier plan du domaine. Nous en avons retiré que la reconnaissance du contexte d’activité peut être obtenue par trois catégories de traitements : par l’activité de l’inférence de base, par l’analyse dynamique de l’activité et par la recommandation d’activités futures. Nous proposons alors un moteur d'inférence intelligent basé sur notre middleware sensible au contexte. Outre les exigences liées à la vérification de la cohérence du contexte, notre moteur d'inférence intègre les trois méthodes les plus populaires concernant la reconnaissance de contexte : des règles, des arbres de décision, et les Modèles de Markov Cachés. Ceci constitue une solution intéressante couvrant toutes les facettes de l'activité de reconnaissance de contexte dans notre middleware sensible au contexte. Les informations collectées à partir des réseaux sociaux sont utilisées pour éduquer le moteur d'inférence intelligent. […] / Almost 20 years ago, Marc Weiser envisioned the prospect of computer in 21st century, and proposed the pioneering notion of ubiquitous computing. One of Weiser’s primary ideas has recently evolved to a more general paradigm known as context awareness, becoming a central research theme in many other ubiquitous computing programs. From Active Badge considered as the first context-aware application, there are numerous attempts to build effective context-aware systems. However, how to acquire context, how to process context and how to create context-aware applications is still faced with enormous challenges in the both of research and practice. This dissertation investigates deeply some chosen key issues in context awareness and develops a context-aware middleware. The main research contributions are presented in three categories: a spatialtemporal context represent model, a context-aware middleware and an intelligence context inference engine. The spatial-temporal context representation model is proposed to organize context and relations for context-aware system. Ontology-based method is adopted to construct our model, supporting both knowledge sharing and reuse as well as logic inference. This model adopts two-layer hierarchy structure for different situation. The higher layer comes up with the generic common context, while the lower layer focuses on various specific situations. Differing from existing models, besides taking locational factors into account, it supports different historical context service depending on different context resource. These context histories may be used to predict and infer the context. A context-aware middleware is designed as a platform associated with context retrieval and context processing. It is organized in two layers: the low layer provides a solution to integrate sensors and actuators with a standardized data representation; the high layer: versatile context interpreter focuses on context processing, which is made up of four parts: Context Aggregator, Inference Engine, Context Knowledge Base, and Query Engine in charge of context inferences, expressive query, and persistent storage. This middleware provides an environment for rapid prototyping of context aware services in ambient intelligent. The intelligent inference engine is the central and intellectual component of context-aware middleware. We review all the methods on activity context recognition published in three premier conferences in past decade and conclude that activity context recognition is divided into three facets: basic activity inference, dynamic activity analysis and future activity recommendation. Then we propose an intelligent inference engine based on our context-aware middleware. Beside satisfying requirements of checking the context consistency, our inference engine integrates the three most popular methods on activity context recognition: Rules, Decision Tree, and Hide Markov Model. It provides a solution for all facets of activity context recognition based on our context-aware middleware. The individuals’ information collecting from their social networks under permission are leveraged to train intelligent inference engine. We finally use two scenarios (applications) to explain the generic process to develop application via our middleware, and compare and analyze the main aspects of our middleware with other five representative context-aware applications. Our middleware profits good features from existing context-aware systems and improve intelligence via supporting activity context recognition. It provides an efficient platform for a rapid developing of new context-aware applications in ambient intelligence.

Ubiquitous Computing can be described as the third stage in the computing history where every user is surrounded by many “computers”. This paper provides an evaluation of a number of multi-agent platforms to decide their appropriateness as an infrastructure for ubiquitous computing. / Ubiquitous Computing kan beskrivas som det 3:e steget i datorns utveckling där varje användare omges med många "datorer". Denna uppsats tillhandahåller en utvärdering av multi-agent platformar för att undersöka deras lämplighet inom Ubiquitous Computing

Trends in technologies have mostly focused on the work environment, entertainment and communication technologies. Some developments have been made for the home such as microwave ovens, washing machines, HDTV, etc but most tasks are still manually executed. Washing machines are used in laundry but there hasn’t been a significant saving in time compared to manual laundering. The effects of misuse of the machine can be very destructive to clothes. This calls for proper sorting of clothes and adjustment of washing settings as appropriate. However, sorting has become a time consuming activity that requires a lot of attention on the part of the individual. Due to fatigue, individuals may in turn not pay much attention to washing instructions. In this thesis, we put together technologies into a system that would aid the user in planning and executing a laundry through identification of dirty clothes and sorting them in groups that can go into separate washes.

Ubiquitous and Pervasive computing put forth a vision where environments are enriched with devices that provide users with services to serve them in their everyday lives. The building of such environments has the final objective of automating tedious routine tasks that users must perform every day. This automation is a very desirable challenge because it can considerably reduce resource consumption and improve users' quality of life by 1) making users' lives more comfortable, eficient, and productive, and 2) helping them to stop worrying and wasting time in performing tasks that need to be done and that they do not enjoy. However, the automation of user tasks is a complicated and delicate matter because it may bother users, interfere in their goals, or even be dangerous. To avoid this, tasks must be automated in a non-intrusive way by attending to users' desires and demands. This is the main goal of this thesis, that is, to automate the routine tasks that users want the way they want them. To achieve this, we propose two models of a high level of abstraction to specify the routines to be automated. These models provide abstract concepts that facilitate the participation of end-users in the model specification. In addition, these models are designed to be machine-processable and precise-enough to be executable models. Thus, we provide a software infrastructure that is capable of automating the specified routines by directly interpreting the models at runtime. Therefore, the routines to be automated are only represented in the models. This makes the models the primary means to understand, interact with, and modify the automated routines. This considerably facilitates the evolution of the routines over time to adapt them to changes in user behaviour. Without this adaptation, the automation of the routines may not only become useless for end-users but may also become a burden on them instead of being a help in their daily life. / Serral Asensio, E. (2011). Automating Routine Tasks in Smart Environments. A Context-aware Model-driven Approach [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/11550 / Palancia

L'association de robots personnels et d’intelligences ambiantes est une nouvelle voie pour l’aide à domicile. Grâce aux appareils intelligents de l'environnement, les robots pourraient fournir un service de haute qualité. Cependant, des verrous existent pour la perception, la cognition et l’action.En effet, une telle association cause des problèmes de variétés, qualités et conflits, engendrant des données hétérogènes et incertaines. Cela complique la perception du contexte et la cognition, i.e. le raisonnement et la prise de décision. La connaissance du contexte est utilisée par le robot pour effectuer des actions. Cependant, il se peut qu’il échoue, à cause de changements de contexte ou par manque de connaissance. Ce qui annule ou retarde son plan. La littérature aborde ces sujets, mais n’offre aucune solution viable et complète. Face à ces verrous, nous avons proposé des contributions, autour à la fois du raisonnement et de l’apprentissage. Nous avons d’abord conçu un outil d'acquisition de contexte qui gère et modélise l’incertitude. Puis, nous avons proposé une technique de détection de situations anormales à partir de données incertaines. Ensuite, un planificateur dynamique, qui considère les changements de contexte, a été proposé. Enfin, nous avons développé une méthode d'apprentissage par renforcement et expérience pour éviter proactivement les échecs.Toutes nos contributions ont été implémentées et validées via simulation ou à l’aide d’un robot dans une plateforme d’espaces intelligents / Personal robots associated with ambient intelligence are an upcoming solution for domestic care. In fact, helped with devices dispatched in the environment, robots could provide a better care to users. However, such robots are encountering challenges of perception, cognition and action.In fact, such an association brings issues of variety, data quality and conflicts, leading to the heterogeneity and uncertainty of data. These are challenges for both perception, i.e. context acquisition, and cognition, i.e. reasoning and decision making. With the knowledge of the context, the robot can intervene through actions. However, it may encounter task failures due to a lack of knowledge or context changes. This causes the robot to cancel or delay its agenda. While the literature addresses those topics, it fails to provide complete solutions. In this thesis, we proposed contributions, exploring both reasoning and learning approaches, to cover the whole spectrum of problems. First, we designed novel context acquisition tool that supports and models uncertainty of data. Secondly, we proposed a cognition technique that detects anomalous situation over uncertain data and takes a decision in accordance. Then, we proposed a dynamic planner that takes into consideration the last context changes. Finally, we designed an experience-based reinforcement learning approach to proactively avoid failures.All our contributions were implemented and validated through simulations and/or with a small robot in a smart home platform

Diploma thesis deals with Ambient intelligence issue. Represents its basic characteristic and demands for its realization. Describes actual stadium in this concept development and shows some present projects. Next focuses on Intelligent buildings issue. In connection with this addresses to so-called human behaviour patterns. Various methods of human behaviour patterns measurement are discussed there. Thesis then focuses on people counting system design, which is based on camera record. Such system represents way of human behaviour patterns measurement. Lastly, the using of this way obtained data is discussed.