The rising ageing population worldwide impacts social and economic facets of modern society, and eventually the quality of human life. There is an increasing demand on a new paradigm of healthcare provisioning that can help address the growing needs of the elderly who are more likely to experience age-related physical frailties and cognitive decline, and also the shortage of public healthcare resources. Ambient assisted living (AAL) with smart homes as a specific realisation of the metaphor has emerged as a realistic technology-driven approach to supporting independent living and delaying institutionalisation. Activity recognition plays a pivotal role in the identification of users' behavioural needs, thus allowing AAL applications to provide context-aware assistive services. This Thesis conceives, designs and develops a hybrid knowledge-driven ontology-based approach for activity modelling and recognition and underlying technologies and tools, which provides a viable technological solution for AAL and advances research frontiers of associated research areas. The Thesis has been conducted through four complementary studies, each addressing a core aspect of the proposed approach. The first study developed and evaluated a hybrid approach to activity modelling that uses ontologies to specify activity models for activities of daily living (ADL), and temporal logic to represent inter-activity relationships for composite activities. The second study developed a dynamic sliding time window-based mechanism for segmenting streaming sensor data to support real-time activity recognition. The mechanism includes a fonnal time window model and its parameters together with algorithms that dynamically manipulate the parameters at runtime to vary the length of the time window. The third study developed a unified approach to simple and composite activity recognition. The approach provided a modular architecture that was realized as a multi-agent system, with agents playing various roles and cooperating to identify simple and composite activities. The final study developed techniques that analyze logged activity data to learn new activities and preferences to adapt initial activity models to make them more complete and responsive. This Thesis implemented ontologies, a software prototype for the proposed approach, and supportive utility tools, e.g. simulator, and synthetic ADL data generator, for experimentation. In addition, all methods and algorithms have been tested and evaluated using various synthetic and real ADL datasets. Experiment results have demonstrated the feasibility of the approach to support real -time activity recognition for both simple and composite activity recognition.

With the inherent functional flexibility of the computer it seemed inevitable that these machines would be put to use in assisting individuals with less functional ability such as the severely physically disabled. These functions would include Speaking, Writing, Controlling their surroundings and moving within those environments. The software applications which provide these functions have collectively become known as Integrated Electronic Assistive Technology (IEAT). This research investigation focused on the prescription of an IEAT prescribed for a particular individual using the Barnsley Wheelchair Interface (BWI) to enable the individual's function of mobility. The BWI controlled an electric wheelchair via a single switch pressed by the user, using her head and was prescribed using the "Multi-Disciplinary Service Model" process created and described by Hawley (1995).The subsequent enhancement of this prescription, from the model-based analysis of the automatically generated usage data logs, that were introduced by the PhD Research student and are one of the novel features of this work. The chosen method, using the analysis of data logs came about as a result of the PhD Research student's own severe physical disability rendering conventional techniques difficult and in some cases impossible. It is the model-based analysis processes that have been developed during this research and which inform the Assistive Technology Professional (ATP) with general information on the prescription, along with comprehensive information on 'Whole System Information', 'Functional Level Information' and 'Interface Level Information', which in all instances had previously been unavailable from assistive technology stakeholders. It is speculated that with the use of these analysis processes (tools) in addition to a software application called "Switch Environment Assessment Tool" (SEAT), in conjunction with the iterative process within the service model, the prescription should be enhanced, resulting in an increase in the user's quality of life.

The new research described in this dissertation created systems and methods to assist wheelchair users and provide them with new realistic and interesting driving opportunities. The work also created and applied novel effort reduction and collision avoidance systems and some new electronic interactive devices. A Scanning Collision Avoidance Device (SCAD) was created that attached to standard powered wheelchairs to help prevent children from driving into things. Initially, mechanical bumpers were used but they made many wheelchairs unwieldy, so a novel system that rotated a single ultra-sonic transducer was created. The SCAD provided wheelchair guidance and assisted with steering. Optical side object detectors were included to cover blind spots and also assist with doorway navigation. A steering lockout mode was also included for training, which stopped the wheelchair from driving towards a detected object. Some drivers did not have sufficient manual dexterity to operate a reverse control. A reverse turn manoeuvring mode was added that applied a sequential reverse and turn function, enabling a driver to escape from a confined situation by operating a single turn control. A new generation of Proportional SCAD was created that operated with proportional control inputs rather than switches and new systems were created to reduce veer, including effort reduction systems. New variable switches were created that provided variable speed control in place of standard digital switches and all that research reduced the number of control actions required by a driver. Finally, some new systems were created to motivate individuals to try new activities. These included a track guided train and an adventure playground that including new interactive systems. The research was initially inspired by the needs of young people at Chailey Heritage, the novel systems provided new and more autonomous driving opportunities for many powered wheelchair users in less structured environments.

Introduction. Les capacités et les possibilités de déplacement en fauteuil roulant manuel (FRM) sont déterminantes pour l'intégration sociale et professionnelle des usagers. Objectifs. Approfondir la connaissance des contraintes de déplacement en FRM à travers 2 études. La 1ère pour identifier les habitudes et préciser les difficultés des usagers ; la 2nde, induite par les résultats de la 1ère, pour évaluer les effets du dévers, encore méconnus, sur les astreintes cardiaques, énergétiques, musculaires et subjectives. Méthodes. La 1ère étude a été menée par questionnaire complété par 118 sujets paraplégiques en FRM. La 2nde étude, en laboratoire avec 25 hommes paraplégiques, comporte 2 tests : un test d?effort sur ergocycle à bras et un test de propulsion de 300 m et 8 conditions combinant 4 dévers (0, 2, 8 et 12 %) et 2 vitesses (0,97 m.s-1 et une vitesse choisie "de confort"). Résultats. La 1ère étude révèle un environnement peu adapté, avec des moyens de transport et des bâtiments souvent inaccessibles sans aide, et des gènes limitant les sorties des usagers parmi lesquelles figure le dévers. La 2nde étude indique des astreintes cardiaques, énergétiques et musculaires significativement modifiées par le dévers et par la vitesse (p < 0,01). L'astreinte subjective n'est pas modifiée par la vitesse. Des différences entre 0 et 2 % sont faibles mais existent. Un dévers de 12 % est difficilement réalisable. Les équilibres des activités musculaires changent dès 2 % de dévers. Le niveau lésionnel modifie de façon très importante les stratégies de déplacements et les niveaux d'astreintes mesurées ou perçues. Discussion-Conclusion. Les effets du dévers sont perceptibles dès 2 % et sont nets et délétères à 8 %. Ce travail permet de proposer de façon plus argumentée un modèle d'astreintes qui conforte le seuil de dévers fixé réglementairement à 2 % permettant à la grande majorité des usagers de FRM de se déplacer de façon autonome et de rester actifs / Introduction. Displacement capacities and possibilities in manual wheelchair (MWC) are strong determinants of social and occupational integration of MWC users. Objectives. Improve the knowledge of the strains involved by displacement in MWC through 2 studies; the first one to identify habits and difficulties of MWC users. The second one, induced by the results of the first one, has assessed the poorly known cardiac, energetic, muscular and subjective strains of the displacement of MWC users on a cross slope. Methods. The first study was a survey answered by 118 paraplegics habitual MWC users. The second one, a laboratory study conducted with 25 paraplegic men, includes 2 tests: a sub maximal arm-cranking test and a 300 m propulsion test with 8 conditions combining 4 cross slopes (0, 2, 8 et 12%) and 2 velocities (0,97 m.s-1 and a "comfortable" chosen one). Results. The first study shows an adverse environment, with many places unreachable without help and disturbances, like cross slope, which hinders users' displacements. The second study demonstrates the increases of cardiac, energetic, muscular and subjective strains with cross slope. Differencies between 0 and 2% cross slopes are low but quantified. Displacement on a 12% cross slope is very difficult. The balance of muscles activities is changing at 2% cross slope. The injury level makes deep changes in the displacements strategies and in as well perceived as measured strain levels. Discussion-Conclusion. Effects of cross slope are perceived for as low as 2% and onwards values. They are high and deleterious at 8%. This work allows to construct a model of cross slope strains which confirms that the regulatory 2% cross slope limit is well chosen because it allows most MWC users to move without help and remain active

Les fauteuils roulants « intelligents » dotés de facultés de navigation autonomes visent à soulager les personnes handicapés moteurs ayant des difficultés à conduire un fauteuil électrique standard. Depuis les années 80, de nombreuses études ont été menées pour réaliser de tels prototypes mais très peu ont abouti à de réels progrès pour les utilisateurs et ce uniquement pour des fonctionnalités simples. Cela peut s'expliquer par plusieurs facteurs, notamment la sous-estimation des contraintes d'acceptabilité et d'adaptabilité. Dans ce contexte, notre objectif est de concevoir une assistance à la conduite sous respect de ces contraintes. Pour cela, nous réalisons un système léger, composé de trois capteurs laser, d'une caméra et d'un micro-ordinateur. Il est susceptible de s'adapter sur tout type de fauteuil électrique et nous a permis de développer les fonctionnalités de franchissement de passage étroit et de suivi de mur ainsi qu'une interface humain-machine ergonomique. Celle-ci est établie par une commande « déictique » qui consiste à fournir un aperçu de l'environnement sur lequel l'utilisateur désigne la tâche autonome choisie par son lieu d'application, le fauteuil réalisant automatiquement l'action correspondante. Tout contact sur le joystick rend immédiatement le contrôle du fauteuil à l'utilisateur afin qu'il ne se sente pas prisonnier de l'assistance, ainsi le pilotage se compose d'une alternance de commandes manuelles et d'indications sur l'interface. Ensuite, l'évaluation de notre assistance à la conduite s'est déroulée en trois étapes. La première, qualitative, a consisté à présenter et à faire essayer le système à des utilisateurs potentiels. Dans un second temps nous avons comparé quantitativement les performances de la conduite assistée avec celles d'une conduite manuelle sur un panel de sujets valides (temps de parcours, nombre d'actions réalisées, indice de confort...). La troisième étape a consisté à évaluer la charge cognitive des utilisateurs dans les deux modes de conduite en mesurant la charge attentionnelle et la capacité décisionnelle à partir d'une méthode de douche tâche / The Smart Wheelchairs, that are equipped for autonomous navigation functionalities, aim to relieve people with disabilities who have difficulty to drive a standard electric wheelchair. Since the 80s, many studies have been conducted to design such prototypes but very few have led to real progress for users and only for simple functionalities. This can be explained by several factors, including the underestimation of the constraints of acceptability and adaptability. In this context, our goal is to provide a driving assistance in respect of these constraints. For this, we design a lightweight system, consisting of three laser sensors, a camera and a computer. It is adaptable to any type of electric wheelchair and allowed us to develop the functionalities of passing through of the narrow passages and wall following, as well as an ergonomic human-machine interface. This latter is established by a deictic command which consists in to provide an overview of the environment where the user indicates the chosen autonomous task by his application location, and then the wheelchair performs the corresponding action automatically. Moreover, all contact with the joystick gives back the control to the user so that he does not feel a prisoner of assistance; well the driving is composed of alternating manual controls and indications on the interface. Finally, the evaluation of our assistance in the conduct took place in three stages. The first, qualitative, is to present and to do try the system to potential users. In a second step, we compare quantitatively the performance of the assistance driving with those of a manual driving on a panel of valid persons (travel time, number of actions, discomfort index?). The third step is to assess the cognitive load of users in both driving modes by measuring the attentional load and decision-making capacity from a dual task method