Self adaptive trajectory learning for monitoring the elderly wandering

Walid Bourennane (1) , Fehd Bettahar (1) , Eric Campo (1)

1 LAAS-S4M - Équipe Instrumentation embarquée et systèmes de surveillance intelligents

Recently, the outdoor and indoor mobility behaviour of elderly has received great attention due to the constant increasing of dependent people. In this paper we describe the architecture of a self adaptive monitoring system which allows detecting cases of wandering or getting lost for people suffering from Alzheimer's disease. The system is composed of Wi-Fi/GPS-based location platform that communicates with a server through a wristwatch. An optimised algorithm embedded on the server processes the stored location data using learning techniques. In result of this operation, safe areas are automatically defined in indoor and outdoor environment according to the usual displacements and time slots. If the person is located out of the safe areas or shows some unusual trajectories, an alert indicating the location of the person is sent by the wristwatch to the medical assistance via GSM/3G service.

Mots clés: Monitoring; elderly; self-adaptive learning; wandering; getting lost; Wi-Fi and GPS location

Domaines: Apprentissage [cs.LG]; Electronique

Walid Bourennane, Fehd Bettahar, Eric Campo. Self adaptive trajectory learning for monitoring the elderly wandering. International Conference on Biomedical Engineering, Feb 2013, Innsbruck, Austria. ⟨hal-02878105⟩