An AMI system for user daily routine recognition and prediction

Abstract

Ambient Intelligence (AmI) defines a scenario involving people living in a smart environment enriched by pervasive sensory devices with the goal of assisting them in a proactive way to satisfy their needs. In a home scenario, an AmI system controls the environment according to a user's lifestyle and daily routine. To achieve this goal, one fundamental task is to recognize the user's activities in order to generate his daily activities profile. In this chapter,we present a simpleAMI system for a home scenario to recognize and predict users' activities.With this predictive capability, it is possible to anticipate their actions and improve their quality of life. Our approach uses a Hidden Markov Model (HMM) to recognize activities and deal with the intrinsic uncertainty of sensory information. The concepts of this domain have been formally defined to allow a higher-level system to enrich its knowledge base. © Springer International Publishing Switzerland 2014.