Context Awareness of Human Motion States Using Accelerometer

Abstract

The proposed context awareness system is composed of acceleration data acquisition part and fuzzy inference system that processes acquired data, distinguishes user motion states and recognizes emergency situations. Two-axial accelerometer embedded in SenseWear PRO2 Armband (BodyMedia) on the right upper arm collects input data containing the longitudinal acceleration average (LAA), the transverse acceleration average (TAA), the longitudinal acceleration-mean of absolute difference (L-MAD), and transverse acceleration mean of absolute difference (T-MAD). Fuzzy inference system is a tool imitating the human ability of decision making. In our system, the fuzzy inference system was used to distinguish the user motion states and to recognize emergency situations. In an experiment using eight subjects, the recognition rates of lying, sitting, walking and running were 98.9%, 98.9%, 99.7% and 99.9%, respectively. Recognition rate for lying after walking and lying after running was 100%.