Development of ZigBee Triaxial Accelerometer based Human Activity Recognition System

Human activity is an important context in ubiquitous computing. This paper presents the hardware and software design and development of a low-cost, wearable and IEEE 802.15.4/ZigBee compliant mote with triaxial accelerometer, Acc-Triax for the monitoring of human physical activities and leg postures. Acc-Triax which is placed in trouser pocket does processing, feature extraction and classification of sitting, standing and walking activities. Customized decision tree is programmed inside the module to classify the activities. 99% accuracy was obtained for sitting, standing and walking activities. In addition to these activities, our algorithm recognizes 11 postures of leg in sitting and standing positions. The activities and leg postures are transmitted wirelessly through ZigBee to the PC/basestation. This system can be used in monitoring of people's activities of daily living (ADL) in an indoor environment. Physical ergonomics of the people may also be monitored from the recognized leg postures.