Behaviour monitoring of the elderly by trajectories analysis

This paper presents a way to keep the elderly autonomous in their daily activities. The principle is to monitor their behaviour and more specifically their trajectories of movement in the living area. Two methods have been developed and compared: supervised and unsupervised classification. Some results from data collected in a long-stay setting are shown. They highlight the possibility of determining normal trajectory classes, and by comparison with the usual situation, to trigger alarms when the trajectories are unusual.