Journal of Ambient Intelligence and Smart Environments

Robotic and ambient intelligent environments are much more demanding regarding the knowledge a robot needs to have. A skilled robot performs human-scale manipulation tasks, interacts with a variety of objects, understands instructions given by humans and most importantly, requires the capability of interpreting ubiquitous resources and assembling them into a complex plan. In this paper, a novel Knowledge-enable Decision Making Framework (KDMF) is proposed for Component-Based Robotic System (CBRS). We exploit that the use of domain knowledge is pivotal to endow robots with higher degrees of autonomy and intelligence in CBRS. Ontology knowledge about classes, properties, relations is organized in OWL based conceptual map, which allows automated inference to derive new pieces of information. Knowledge about tasks is specified in a tree data structure, and knowledge about components' functions is formulated by a specific type of service specification profile. Using the knowledge representation, a task-oriented decision making method is proposed that integrates knowledge inference and service components utilization. In practical applications of ambient and robotic assisted living, robot's plans generated by the decision making software are based on the knowledge of components, rather than particular device instances, which improves the reusability and flexibility of the system. Experimental results validate the effectiveness of the proposed method.

Inhabitants of today's smarter homes struggle with complicated user interfaces and inflexible home configurations. The proposed smart home recommender system addresses these issues by continuously interpreting the user's current situation and recommending services that fit the user's habits, i.e. automate some action that the user would want to perform anyway. With these recommendations it is possible to build much simpler user interfaces that highlight the most interesting choices currently available. Configuration becomes much more flexible, since the recommender system automatically learns user habits. Evaluations on two smart home datasets show that the algorithm produces correct recommendations with 61% and 73% accuracy, respectively.

As the number of elderly people in our society increases, the need of assistive technologies in home becomes urgent. Existing techniques allow elderly people to be better assisted through monitoring what goes on in smart homes and inferring their activities from sensor data via a recognition model. However, there are various cases that existing models have difficulties in accommodating relational data. In this paper, we present an application of probabilistic graphical model – Latent-Dynamic Conditional Random Field – to detect the goals of the individual subjects when observations have long range dependencies or multiple overlapping features. To validate the proposed method, we apply it to recognize activities in two different datasets which were collected in smart homes. The results demonstrate that Latent-Dynamic Conditional Random Fields favorably outperform other models, especially when there are extrinsic dynamic activities changes and intrinsic actions (subactivities).

We are often, consciously or unconsciously, self-assessing our quality of life in order to make decisions about our future actions. People with special needs are sometimes not able to perform this evaluation, this being the responsibility of their relatives or carers. The literature shows this to be a challenging task due to the inherent subjectivity, and the limited data collection tools and biased information available. This paper proposes that context awareness and artificial intelligence can support this task by providing digested and objective information about a person's quality of life evolution. Ambient Assisted Living continuously obtains relevant data from different sources such as sensors, the use of household appliances and interaction with user interfaces. An artificial neural network model known as self-organizing maps processes this data to monitor how the user carries out different activities of daily living (e.g. cooking or doing the washing). This information, together with statistical analysis from the said data, is automatically compiled by the system in a report to visualize trends in user behavior that might lead to the detection of a person's cognitive, physical or sensory deterioration. This report has been validated by a group of experts who considered it a tool of great usefulness and power to complement existing tools used by social workers.

In the last few years, with the support of new communications and mobile technologies, Physical Annotations (PAs) have become popular in context-aware systems and pervasive computing. PA aims to enrich physical entities around us with useful information. Users can interact with inanimate objects, annotate them and share annotations with others. It helps daily life in different aspects, such as tourism, education and health. In this paper, we provide a formal model and theoretical concepts for PA systems. We also propose a new location model called DPVW-model for annotated entities. DPVW-model represents aspects of the annotatable physical world as well as the virtual domain. Then, we discuss the implementation and usability study of our system. Finally, we compare it with other similar systems.

In this paper, we introduce a novel method for view-independent hand pose recognition from depth data. The proposed approach, which does not rely on color information, provides an estimation of the shape and orientation of the user's hand without constraining him/her to maintain a fixed position in the 3D space. We use principal component analysis to estimate the hand orientation in space, Flusser moment invariants as image features and two SVM-RBF classifiers for visual recognition. Moreover, we describe a novel weighting method that takes advantage of the orientation and velocity of the user's hand to assign a score to each hand shape hypothesis. The complete processing chain is described and evaluated in terms of real-time performance and classification accuracy. As a case study, it has also been integrated into a touchless interface for 3D medical visualization, which allows users to manipulate 3D anatomical parts with up to six degrees of freedom. Furthermore, the paper discusses the results of a user study aimed at assessing if using hand velocity as an indicator of the user's intentionality in changing hand posture results in an overall gain in the classification accuracy. The experimental results show that, especially in the presence of out-of-plane rotations of the hand, the introduction of the velocity-based weighting method produces a significant increase in the pose recognition accuracy.