Ubiquitous sensors based human behavior modeling and recognition using a spatio-temporal representation of user states

There has been much research on user activity assistance applications using the location of users and objects as context. However people's activities are described in terms of time sequence aspect in addition to location aspect. Therefore, it is important for enhanced user activity support systems to consider the user's context in terms of spatio-temporal constraints. In this paper, we propose a user activity assistance system that employs a state sequence description scheme to describe the user's contexts. In this scheme, each state is described as a spatio-temporal relationship between the user and objects in the real world. Typical sets of states are stored as models of tasks performed by a user. To try out this system, we have developed an experimental house containing various embedded sensors and RFID-tagged objects. Each state is detected by a decision tree constructed by a C4.5 algorithm using the output of the sensors and the RFID tags. The user's context is obtained by matching the detected state series to a task model. Having evaluated the performance of the proposed system in this experimental house, we conclude that our system is an effective way of acquiring the user's spatio-temporal context.