Simple rule-based human activity detection with use of mobile phone sensors

Abstract

The human activity recognition with use of new generation of smartphones equipped with various sensors became widely used technique lately. Many works concern especially one type of incidents-fall detection. This paper concerns detection of different activities. The proposed approach is based on use of simple rules that make possible to distinct considered events. The rules are defined with use of measured acceleration and phone orientation and take into account moments of event occurrences. The values of parameters used in rules were defined by experiment performed with use of implemented application on Android system. Finally, the accuracy of presented method is examined in the paper.