A Living Lab Infrastructure for Investigating Activity Monitoring Needs in Service Robot Applications

This paper presents a framework that has been developed for automatic activity recognition and domestic behavior monitoring, towards supporting elderly MCI patients in their daily domestic life. Our framework's infrastructure consists of a network of smart-home sensors and RGB-D cameras that can be adapted and be unobtrusively installed in a variety of indoor living areas, collecting data relative to the human's movement and the state of the home environment. User activities and behavior are then assessed through machine learning algorithms applied on these data. The developed framework has been applied in real house settings and extensive analysis has been performed, so as to investigate how human activity and behavior monitoring needs, in the scope of ICT solutions for supporting active and healthy ageing of MCI patients, can be covered in the scope of corresponding service robot applications.