Human activity recognition using a fuzzy inference system

Abstract

This paper presents a fuzzy inference system (FIS) for recognizing human activities using a triaxial accelerometer. The accelerometer is used to collect human motion acceleration data for classifying four different activities: moving forward, jumping, going upstairs, and going downstairs. Three different features including peak to peak amplitude, standard deviation, and correlation between axes are extracted from each axis of the accelerometer as inputs to the fuzzy system. The fuzzy rules and the membership functions of this fuzzy system are defined based on the experimental values of these features. The experiments show that the proposed fuzzy inference system recognizes moving forward, jumping, going upstairs, and going downstairs with accuracy of 100%, 96.7%, 93.3%, and 93.3%, respectively.