An IoT based Virtual Coaching System (VSC) for Assisting Activities of Daily Life

Nowadays aging of the population is becoming one of the main concerns of theworld. It is estimated that the number of people aged over 65 will increase from 461million to 2 billion in 2050. This substantial increment in the elderly population willhave significant consequences in the social and health care system. Therefore, in thecontext of Ambient Intelligence (AmI), the Ambient Assisted Living (AAL) has beenemerging as a new research area to address problems related to the aging of the population. AAL technologies based on embedded devices have demonstrated to be effectivein alleviating the social- and health-care issues related to the continuous growing of theaverage age of the population. Many smart applications, devices and systems have beendeveloped to monitor the health status of elderly, substitute them in the accomplishment of activities of the daily life (especially in presence of some impairment or disability),alert their caregivers in case of necessity and help them in recognizing risky situations.Such assistive technologies basically rely on the communication and interaction be-tween body sensors, smart environments and smart devices. However, in such contextless effort has been spent in designing smart solutions for empowering and supportingthe self-efficacy of people with neurodegenerative diseases and elderly in general. Thisthesis fills in the gap by presenting a low-cost, non intrusive, and ubiquitous VirtualCoaching System (VCS) to support people in the acquisition of new behaviors (e.g.,taking pills, drinking water, finding the right key, avoiding motor blocks) necessary tocope with needs derived from a change in their health status and a degradation of theircognitive capabilities as they age. VCS is based on the concept of extended mind intro-duced by Clark and Chalmers in 1998. They proposed the idea that objects within theenvironment function as a part of the mind. In my revisiting of the concept of extendedmind, the VCS is composed of a set of smart objects that exploit the Internet of Things(IoT) technology and machine learning-based algorithms, in order to identify the needsof the users and react accordingly. In particular, the system exploits smart tags to trans-form objects commonly used by people (e.g., pillbox, bottle of water, keys) into smartobjects, it monitors their usage according to their needs, and it incrementally guidesthem in the acquisition of new behaviors related to their needs. To implement VCS, thisthesis explores different research directions and challenges. First of all, it addresses thedefinition of a ubiquitous, non-invasive and low-cost indoor monitoring architecture byexploiting the IoT paradigm. Secondly, it deals with the necessity of developing solu-tions for implementing coaching actions and consequently monitoring human activitiesby analyzing the interaction between people and smart objects. Finally, it focuses on the design of low-cost localization systems for indoor environment, since knowing theposition of a person provides VCS with essential information to acquire information onperformed activities and to prevent risky situations. In the end, the outcomes of theseresearch directions have been integrated into a healthcare application scenario to imple-ment a wearable system that prevents freezing of gait in people affected by Parkinson’sDisease.