A Body Activity Tracking System using Wearable Accelerometers

A body activity tracking system based on wearable accelerometers is presented in this paper. The system maintains multiple models for tracking different movements and employs a selection strategy to choose a suitable model. The measured accelerations are converted into flexion angles of body segments after such processing. The system is applied to track ambulatory activities and the result shows that its performance is comparable to that of vision-based and magnetic sensor based approaches in tracking accuracy and smoothness while the computation complexity is very much lessened.