One of the challenges of the ageing population in many countries is the efficient delivery of health and care services, which is further complicated by the increase in neurological conditions among the elderly due to rising life expec‑ tancy. Personal care of the elderly is of concern to their relatives, in case they are alone in their homes and unforeseen circumstances occur, affecting their wellbeing. The alternative; i.e. care in nursing homes or hospitals is costly and increases further if specialized care is mobilized to patients’ place of residence. Enabling technologies for independentliving by the elderly such as the ambientassistedlivingsystems (AALS) are seen as essential to enhancing care in a cost‑effective manner. In light of significant advances in telecommunication, computing and sensor miniaturization, as well as the ubiquity of mobile and connected devices embodying the concept of the Internet of Things (IoT), end‑ to‑end solutions for ambientassistedliving have become a reality. The premise of such applications is the continuous and most often real‑time monitoring of the environment and occupant behavior using an event‑driven intelligent system, thereby providing a facility for monitoring and assessment, and triggering assistance as and when needed. As a growing area of research, it is essential to investigate the approaches for developing AALS in literature to identify current practices and directions for future research. This paper is, therefore, aimed at a comprehensive and critical review of the frameworks and sensor systems used in various ambientassistedlivingsystems, as well as their objec‑ tives and relationships with care and clinical systems. Findings from our work suggest that most frameworks focused on activity monitoring for assessing immediate risks, while the opportunities for integrating environmental factors for analytics and decision‑making, in particular for the long‑term care were often overlooked. The potential for wearable devices and sensors, as well as distributed storage and access (e.g. cloud) are yet to be fully appreciated. There is a distinct lack of strong supporting clinical evidence from the implemented technologies. Socio‑cultural aspects such as divergence among groups, acceptability and usability of AALS were also overlooked. Future systems need to look into the issues of privacy and cyber security.
The combination of an aging society and the progress of digitization offer both opportunities and risks related to new technology. These technologies should meet the requirements of people who have lived without smart phones, tablet computers and the internet for decades. In particular, in order to be useful, new functionalities should offer a noticeable added value for older people in everyday life. Due to the growing demand for health services, which are especially needed for the elderly, increasing health care expenditures are anticipated (Bundesministerium des Innern, 2011). The lack of qualified caregivers and decrease in family care (Geiger, 2009) present an opportunity to use assistive technology in the residential environment. Technical support via comfort functions, such as regulating the temperature in the apartment or operating lights, has been offered by various vendors in Germany in previous years. By adding more sensors that can measure physiological status (e.g. vital functions such as blood pressure, or automatically alerting the emergency doctor in the event of a fall), AAL systems can provide additional security for people with health problems or physical impairments. With the exception of the control element, other AAL components and services are often intangible or function only in the background. Thus, these components could be difficult to understand for the end-user. The research project Smart and IndependentLiving
• Heterogeneous an hierarchical Augmenting a space usu- ally involves the integration and coordination of sev- eral different kind of system nodes and devices. A hierarchical organization of them can help in building a richer encapsulated functionality. For example, a sensor node can provide raw sensorial data, a sensor network on the other hand, can constitute a sort of “macroscope” we can use to identify composite activi- ties, while hiding us the individual nodes coordination. • Context-aware. Three steps are involved: first, raw context data such as temperature, identity, localiza- tion, and others, is acquired; second, context is mod- eled, communicated through the system, and then in- terpreted; third, context-inferred data is used to per- form an action or cast a change in the system. This requirement becomes really important for the elder be- cause some systems usually rely on personal context such as blood pressure, sugar levels, and others. Con- text in such situations should be communicated in fast, reliable and opportune ways.
Abstract—The recent increase in ageing population in countries around the world has brought a lot of attention toward research and development of ambientassistedliving (AAL) systems. These systems should be inexpensive to be installed in elderly homes, protecting their privacy and more importantly being non-invasive and smart. In this paper, we introduce an inexpensive system that utilises off-the-shelf sensor to grab RGB-D data. This data is then fed into different learning algorithms for classification different activity types. We achieve a very good success rate (99.9%) for human activity recognition (HAR) with the help of light-weighted and fast random forests (RF).
To support AAL, different location and tracking technologies are being exploited in many AAL systems. Technologies like WiFi, Global Positioning Systems (GPS), Ultra Wide Band (UWB), ultrasound, Camera-based, and Infrared Radiation (IR) have been widely used for tracking people inside/outside houses[5-7]. Based on these technologies, various commercial and experimental systems have been developed such as Ekahau , Microsoft RADAR , Active bat , Active badge , and Ubisense . Most of these systems use some sort of wearable devices such as wearable tags, bracelets, pendants, or sensors like cameras, and motion sensors to detect peoples’ locations. Many other technologies are also being integrated into advanced AAL systems, such as wearable sensors to monitor vital signs, or teleconferencing to combat loneliness, but these other technologies are outside the scope of this paper. Even though the existing location and tracking services are considered a great step-forward for providing location knowledge, they are not quite adequate for AAL. The design principles of most of these services are not fully matched with all the requirements of AAL. For instance, most of these services require a considerable number of location sensors/devices to be deployed in the environment in order to facilitate the collection of location data (e.g. RFID and ultrasound technologies). This, however, has led to a non-trivial increase in the installation effort and the overall cost for deploying the service. AAL services are mostly intended to be used in houses/residential environments by non-technical elderly
The problem of detecting unusual changes in the daily behaviour of an elderly person who lives independently at home has been widely investigated in the literature [ 1 ]. Solutions typically are sensor-based systems that require the use of wearable and non-wearable sensors to track the daily behaviour and provide responses when deviations are detected. These solutions usually require the intervention of the resident user, for instance, by pushing a button on a pendant or on a wrist watch, or by monitoring the resident using camera-based sensors installed at different locations in the house. However, prior research studies [ 2 ] show that wearable and camera-based sensors are not very appreciated by the elderly people due to inconvenience, are computationally complex, and raise privacy issues. The elderly might not feel comfortable wearing sensors all the time and may forget to wear them on some occasions, or may feel they are losing their privacy when being monitored by cameras at home. This reduces the usefulness of these sensors for continuous behavioural monitoring. Even though some recent research studies propose including the sensors in people’s clothes [ 3 ] or utilising the capabilities of smart watches [ 4 ] for behaviour monitoring, these works are still limited
This paper reports the usability evaluation of interfaces developed to enable elderly and disabled people interact remotely with kitchen appliances in the home to enhance their independentliving. A number of evaluation exercises were undertaken throughout the project’s development, including user-participative workshops and focus groups. This paper focuses on the summative usability evaluation exercise, which comprised a laboratory-based study in a simulated home environment, with a view to determining the appropriateness of employing this approach with potentially vulnerable participants. The study involved 27 participants interacting with the user interface. Their behaviour was observed and recorded, and their interaction with the system was analysed. They were also given a post-session questionnaire, where their opinions of the usability of the interface were solicited. The results of the usability testing were positive, and insight has been gained into how products of this nature can be further improved. The experience of conducting laboratory-based studies with vulnerable users was positive and led to propose in this paper a set of guidelines for future work in evaluating usability for work in this domain.
Smart home systems generally exploit ambient sensors to measure physical quantities (e.g., temperature, lighting, and humidity), and actuators to automatically influence them depending on users' preferences . Additionally, AAL systems leverage wearable sensors installed in smartphones, smart watches, and wristbands, to better monitor the user's behavior and disambiguate among similar situations . In case of specific health conditions, it is also possible to accurately monitor patients' vital signs by means of specialized devices . For a survey on AAL tools to improve elderly people’s lives, please refer to . However, accurately fusing data coming from such heterogeneous sensors can prove to be very challenging, given the intrinsic inaccuracies of sensory measurements . In this regard, probabilistic techniques, such as Dynamic Bayesian Networks (DBNs) , can help by explicitly modeling the uncertainty of data collection. In recent years, DBNs have been exploited by many state-of-the-art systems for different applications, such as fire detection , target tracking , and user presence detection .
Authors disagree on the functionalities of ICT devices, with some proposing that they should be reduced (Wallace et al. 2010) and others proposing that they should be ex- tended by providing more alarms and more functions (Zaad and Ben Allouch 2008). It should be borne in mind that the elderly (even those with mild dementia) are still able to learn, albeit in a different way than usual (Wallace et al. 2010). Nevertheless, overly complex systems with multiple-step procedures that place high learning requirements on the diminished capabilities of PwDs have a greater likelihood of failure (O’Neill, Ma- son, et al. 2011; Wallace et al. 2010). For instance, blinking LEDs or vibrating sounds (J. van Hoof et al. 2011), a screw head looking like a button (O’Neill, Mason, et al. 2011), or an amount of newly installed cables (J. van Hoof et al. 2011) can cause confusion or frus- tration and can also have a major impact on the overall acceptance of the technology for the user. PwDs will also have to learn to cope with AAL technologies, when employing these assistive tools (Portet et al. 2011). A further way of enhancing the acceptability of a system for the user is by the provision of sufficient customisation, adaptation possibil- ities or high quality products (Abascal and Azevedo 2007; Francis et al. 2009).
Apart from AAL systems, the use of assistive robots as an aid in elderly daily lives is also an active research topic widely explored [15–17]. A good revision of robot assistants can be found in . Some of them have been tested in laboratory scenarios, like Nao robot in , but in the most interesting proposals, the robots have been tested in real homes, even in long-term experiments. Hobbit robot  is a relevant case study. It is a care robot which is capable of fall prevention and detection as well as emergency detection and handling. Its interaction with the user is based on a multimodal user interface including automatic speech recognition, text-to- speech, gesture recognition, and a graphical touch-based user interface. Other interesting examples are the Max companion robot , several robotic platforms developed inside the EU FP7 Robot-Era Project , the GIRAFF robot , and the telepresence robot in .
the tested device: the GP-m process cannot handle such a high number of requests per second, that saturate the Binder transaction buffer (that has a limited fixed size (1Mb) and it is shared by all transactions in progress for the process). This is reasonable especially because in AAL application the system should react with timings compatible with the user, which hardly requires high frequencies . Note however that, if higher frequency in data sampling is required, the actual rate of communication is generally lower due to the need for data aggregation and fusion techniques . In further tests on scalability we have considered frequencies up to 5 rps for a producer service. Figure 3.15 shows the scalability of GP-m in terms of number of concurrent producers on the same device. We identified a limit of 40 producers at 5 rps with 100B of message payload. Also in this case the limitation is due to the hardware of the tested device for the same reason of the previous case. Finally, Figure 3.16 shows the scalability of GP-m in terms of number of concurrent consumers on the same device varying the number of concurrent producers trans- mitting at 5 rps with 100B of message payload. Each consumer is subscribed to all the present producers in the network. The GP-m middleware can handle, with an aggregated latency under 30 ms, up to 10 concurrent consumers in presence of 50 producers, up to 20 consumers when 25 producers are transmitting simultaneously, up to 30 consumers with 10 concurrent producers, and more than 100 consumers subscribed to a single producer. We tested the GP-m middleware in particular stress conditions with the mobile device acting as a single aggregation point of all the pos- sible services installed in the home environment. We tested GP-m fully integrated in a GiraffPlus system installed in 15 real homes (5 in Sweden, 5 in Spain, and 5 in Italy respectively) where the system has been deployed and used for one entire year. We tested the context-awareness of GP-m developing a concrete application that turn off unnecessary network adapters like Bluetooth or WiFi once it detects an out- side scenario. We believe that such an application is very useful to limit the energy consumption of mobile devices exploiting the context data coming from the Giraff- Plus network. Elderly people living alone in their house is a scenario particularly dear to the AAL community. Also in the GiraffPlus test sites, elderly people live alone, so when no presence sensor data is sent on the context buses, we can infer that the user is outside the house. To estimate the increased battery saving of this solution, we compared the power consumption of the context-aware application built upon GP-m with a generic application that periodically scans the WiFi signal strength to fingerprint the home WLAN. We also compared our solution with a generic situation where the user forgets all the used network interfaces on when he goes outside (ALL-ON ).
There are many other studies in the subsystem areas of our AAL system. The related works are examined for each of the detection systems one by one. As a start, a survey paper about principles and approaches for the falling person detection is published last year . It is accepted that fall detection systems help elderly people to maintain an independent way of life , . Currently, the commercially available fall detection systems feature body-worn sensors which are connected to a wireless network , . There are other recent studies which turn a mobile phone into a fall detector . Computer vision and Kinect sensor based fall detectors are studied in various works , , . Vibration sensor based fall detection systems are proposed in two studies so far , . However, pattern recognition algorithms for fall detection are not described in the mentioned articles. Without a recognition algorithm the resulting system will not be a robust and practical system. There are also multi-PIR-sensor based fall detectors which use generally more than four PIR sensors , .
When people are getting old, a relevant source of frustration comes from losing physical strength, but what torches them most lies probably in the psycho- logical sphere: they are becoming passive consumers of the societal services rather than active producers. In so doing, they also lose their self-esteem. Almost all the AAL systems for the elderly people consider their users as people who are weak and require to be passively assisted by others. For the designers of such systems, being able to maintain some degree of independence without bringing too much burden to our society appears as an already ambitious goal. However, those systems neglect the fact that the el- derly people can still make their contributions to our society through their valuable experiences. A home- care system with human participation could help to encourage the elderly people to actively participate in group activities as peer participants, and possibly even to use their experiences to help the younger generations to solve, e.g., some of their work and school problems . We expect these activities could help the elderly people to find themselves still useful and thus enable them to live in more active ways. Such possibilities will be discussed in more details in the following section.
12. Inoue, K.; Wada, K.; Uehara, R. How Effective Is Robot Therapy?: PARO and People with Dementia. In Proceedings of the 5th European Conference of the International Federation for Medical and Biological Engineering, Budapest, Hungary, 14–18 September 2011.
13. Coradeschi, S.; Cesta, A.; Cortellessa, G.; Coraci, L.; Galindo, C.; Gonzalez, J.; Karlsson, L.; Forsberg, A.; Frennert, S.; Furfari, F.; et al. GiraffPlus: A System for Monitoring Activities and Physiological Parameters and Promoting Social Interaction for Elderly. In Human-Computer Systems Interaction: Backgrounds and Applications 3; Hippe, Z.S., Kulikowski, J.L., Mroczek, T., Wtorek, J., Eds.; Springer International Publishing: Cham, Switzerland, 2014; pp. 261–271. [ CrossRef ]
Abstract. The ageing process of population of the First World coun- tries is increasing the interest towards solutions to improve the quality of life of elderly or disabled people and their families, providing an econom- ically sustainable healthcare. Thus AmbientAssistedLiving (AAL) and Ambient Intelligence (AmI) are moving towards the Artificial Intelligence field to develop modular, adaptable and intelligent systems to cope with changing needs that characterize the life of people with chronic diseases. These systems should find smart and simple ways to improve the patient quality of life, enhancing at the same time the communication between the system, the assisted person, his family and the medical staff, acting also as a filter of information that provides useful and significant data. This paper introduces the basic architecture of an expert system for AAL: the Virtual Carer (VC). It has to be understood as an IT system modelling a distributed, reliable and modular sensor network composed by biometric and ambient sensors, being able to communicate with an assisted person, to monitor his health conditions and to control the en- vironment around him. The main goal of the system is to help an elderly patient with his daily activities ensuring his security. The proposed sys- tem is based on a multi-agent architecture to ensure its flexibility and interoperability: new devices or sensors can be added simply adding new agents, thanks to the standardization of agent communication. The sys- tem includes also a reasoning part based on Belief-Desire-Intention (BDI) paradigm, trying to model the behaviours of a human caregiver.
in which the groups can engage, helped overcome these problems.
The APEX prototyping approach was used there- fore as the medium of communication in a participa- tory design process for a proposed AAL in a care home for the elderly. The focus was a “concern for the user’s point of view” (Halskov and Hansen, 2015). A virtual environment, with connected physical devices, was used to enable participants to explore design ideas, to explore their needs and to contribute suggestions for redesign. The prototype provided a vivid and appropriate expe- rience for participants. They were sufficiently immersed in the proposed environment that it was as if they were there. For example, one participant expressed concern about her privacy when other participants began to en- ter her room in the virtual environment while explor- ing a scenario. The environment was not threatening. It simply extended the kind of experience they were already used to while watching television.
Abstract—As the population in many countries is steadily aging, allowing elderly people to stay longer at home is a growing concern. AmbientAssistedLiving (AAL) proposes new techniques to help people remain autonomous, based on ambient intelligence. We present an ontology-based framework in which ontologies enable the expression of users’ preferences in order to personalize the system behavior. They are also used for the discovery and interconnection of devices, the storage and retrieval of collected data and the transmission of actions. Basing everything on ontologies allows the designer to express the behavior of the system using high-level logic rules. To render AAL systems as autonomous as possible, devices that fail should be detected at runtime. For this reason, the framework offers a diagnosis service that builds a prediction model of the values detected by sensors. It is based on information discovered opportunistically at run-time and knowledge about physical laws. The framework monitors the run-time behavior of the AAL system and uses the prediction model to detect inconsistencies and hence faults. Therefore, fault detection is totally dynamic and opportunistic; there are no pre-defined control loops. This paper describes an actual implementation, with precise technological details, in order to prove the feasibility of the technical choices, and provide implementation ideas for future projects.
Reasoning and planning are intersective and conjunctive issues in the study of AAL system. The role of reasoning in context aware systems is to deduce and derive from the basic context information (includes information unknown, ambiguous, imprecise, and erroneous) to generate mean- ingful information and support system decision making. Planning concerns the problem of how to achieve a goal state starting from a known initial state. To achieve a goal, the system needs to deduce the existing knowledge based on the available context data. An entire process of planing produces a sequence or partially ordered collection of actions that if executed starting from the initial state, which is expected to achieve the goal state. There are several ways that planning can be used in AAL scenarios, for example in an AAL system, planning can be used to coordinate the capabilities of the available resources to provide a solution or perform a task; planning for AAL may have to deal with multiple agency; planners can be used, for example, to schedule task for speciﬁc status. Research in the area of AI planning has made notable progress over the last decade. There are many state-of-art reasoning and planning algorithms have impacted different application areas for AAL according to the surveys by [147 – 149] .
On not yet analyzing specific evaluation results, it is noticeable how caregivers give much less importance than end users to the use of the kitchen (show information/program appliances through adapted user interface); something that a priori would help end users using white goods for more time and thus remaining independent longer. Everybody coincides in the importance of alarm detection and handling in case there is no response. And carers also bestow great importance to “Detect routine changes in the kitchen to inform whenever there are changes in patterns of conduct that can identify any loss of abilities in the user” as it can strongly improve the tools that they have to monitor the progress of elderly and disabled people.
The problem of detecting unusual changes in the daily behaviour of an elderly person who lives independently at home has been investigated widely in the literature . Solutions typically are sensor-based systems that require the use of wearable and non-wearable sensors to track the daily behaviours and provide responses when deviations are detected. These solutions usually require the intervention of the resident user, for instance, by pushing a button on a pendant or on a wrist watch, or by monitoring the resident using camera-based sensors installed at different locations at home. However, prior research studies  show that wearable and camera-based sensors are not very appreciated by the elderly people due to inconvenience, computational complexity, and privacy issues. The elderly might not feel comfortable wearing sensors all the time and may forget to wear them on some occasions, or may feel losing their privacy when monitored by cameras at home. This reduces the usefulness of these sensors for continuous behaviour monitoring. Even though some recent research studies include the sensors into the people clothes  or utilise the capability of smart watches  for behaviour monitoring, these works are still limited and not affordable for everyone, besides the limitations of the sensors’ battery energy. Moreover, most of the existing systems entail an explicit annotation or labelling process to be made offline in order to manually configure the typical behaviour of the monitored persons before use, which increases the required installation time of these systems and prevents them from being adaptive to small shifts in behaviour that do not necessarily should be considered unusual behaviour (e.g. seasonal changes).