A Ubiquitous Care-Support Service based on Agent-oriented Multiple-Context Coordination

Abstract

This paper presents a gentle system for supervising care-support services that fulfill users' actual requirements based on their physical locations and statuses of system components in ubiquitous computing environments. To address issues of traditional supervisory systems based only on location information and the situation of one side of the watching site or the watched site, we consider all the detailed situations (contexts) of associated entities such as devices, software, networks, and users, in both sides, in addition to the users' physical locations. We propose a ubiquitous supervision system called uEyes to realize this. We introduce an autonomous decision making ability and cooperative behavior to each entity in uEyes, based on agentbased computing technologies. Using advanced features of the entities, live video streaming systems for watching over people can be constructed autonomously according to the multiple contexts of the entities, on both sides in runtime. We implemented a prototype system of uEyes for watching over elderly people, and performed some experiments based on several scenarios. For instance, we assumed a scenario in which a son, who is caring for his ailing father, needs to see his father's facial color and expression with high-quality video. For that purpose, a live video streaming system involving a high-resolution camera and a display device was autonomously configured in run-time. We confirmed that supervision services that fulfill detailed users' requirements can be provided effectively.