Activity Recognition in Ubiquitous Computing Environment

摘要

Activity recognition is one of research issues in ubiquitous computing. In this paper we present an approach to recognize complex activity given the formal context representation. Complex activity refers to the activity composed of simple actions to achieve a certain goal, and the formal context representation is inferred context derived from sensor data. We first propose activity model, into which the relations between activity and context patterns are integrated. The relations are expressed by a set of temporal constraints on context. Then the recognition algorithm, which is based on the activity model and activity flow, is described. We further interpret the approach by the case of single-crystal X-ray diffraction experiment.