Human Presence Detection to Support Contextual Awareness in Ambient Assisted Living Scenarios

Abstract

Assistive technologies and ambient assisted living (AAL) environments promote independence and safety at home, especially to vulnerable users such as older adults or people who are recovering after a hospital stay. To support these technologies, we present an approach to detect the presence of people in individual locations of the University of Hertfordshire’s Robot House, a four-bedroom residential house with smart sensors and robots. Specifically, our method provides contextual information to assistive services, enabling tailored support based on the specific location within the home. We assess the combined affordances of a series of low-resolution sensors in contributing to the ambient assisted living scenarios as an active part of a pipeline dedicated to developing personalised service provision at home. Moreover, we used lower-level features and combined sensory data to identify activities of daily living and gain insights into residents’ habits. Our studies reveal that combining two or more sensors contributes significantly to the accuracy of presence detection, as individual sensors can lead to incomplete or biased information. The information we derive from a combination of sensors can be beneficial when ambient assistive technologies are used in the context of virtual wards to tailor a proactive, personalisable and predictive AI-powered observation deck to support patients in their homes.