POSSIBILISTIC ACTIVITY RECOGNITION IN SMART HOMES FOR COGNITIVELY IMPAIRED PEOPLE

Abstract

In order to maintain and improve the quality of life in the homes of people with dementia, we must elaborate assistive technologies that will alleviate the effects of cognitive decline, such as those erroneous behaviors associated with carrying on activities of daily living (ADL). In order to provide adequate assistive services at the opportune moment, it is necessary to recognize the occupant behavior from sensor events of the smart home. To address this challenge, this paper presents a formal model of activity recognition in smart homes based on possibility theory and environmental contexts. This activity recognition model is able to take into account coherent and erroneous behaviors that may result from the cognitive decline of the occupant. An implementation of this model was validated in a smart home laboratory with valid sensor data from coherent and erroneous realizations of an ADL scenario.