Towards application of non-invasive environmental sensors for risks and activity detection

One of the main goals of Ambient Assisted Living (AAL) is to provide supportive environment for the elderly or disabled. Such environments are not feasible without correctly identifying states and activities of the persons receiving the care. They rely on the interaction and processing of data originating from many components and objects in the surrounding. In order to collect the data, various sensors are used to monitor the environment, as well as the person's health parameters. One of the main concerns in AAL is preservation of user's privacy. In this paper we address that by proposing a non-intrusive approach for data collection and identification of daily activity and risks. We describe the wiring of such system based on cheap non-intrusive sensors, deployment in a real environment, the protocols for data fusion and processing, and explain how machine learning could be employed for detecting risks and activities. The main contribution of this paper is development of non-intrusive sensor kits that can be easily deployed in real-life environments and are capable of collecting data that can reliable detect activities and risk.