Distinguishing activities of daily living in a multi-occupancy environment

Abstract

Ambient sensor systems in an intelligent environment are often used to monitor and infer Activities of Daily Living (ADL). Existing research on the recognition of ADL from ambient sensors assumes that the intelligent environment is inhabited by a single-occupant (older adults). However, in the real environment there may be situations in which there is more than one occupant for a time. The focus of this study is to distinguish the number of people in the home environment by employing only PIR sensors in order to identify whether the environment is used by one person or more. In this paper, two different techniques, Fuzzy Entropy (FuzzyEn) and Indoor Mobility (IM), are investigated to distinguish the activities within a multi-occupancy environment. The model is tested and evaluated based on a set of data representing a multi-occupancy environment. The experiment results show that FuzzyEn and IM can detect and identify the existence of a visitor in a home environment with an accuracy of 100% and 98.7%, respectively.