Ambulatory monitoring of human posture and walking speed using wearable accelerometer sensors.

This paper describes a new classification system for real-time monitoring of physical activity, which is able to detect body postures (lying, sitting, and standing) and walking speed with data acquired from three wearable biaxial accelerometer sensors deployed in a wireless body sensor network. One sensor is waist-mounted while the remaining two are attached to the respective thighs. Two studies were conducted for the evaluation of the system, with each study involving five human subjects. Results from the first study indicated an overall accuracy of 100% for classification of lying, sitting, standing, and walking across a series of 40 randomly chosen tasks. In our system, estimated walking speeds are used to distinguish between different types of movement activity (walking, jogging, and running), and the accuracy of its estimation was evaluated in our second study which gave an overall mean-square error (MSE) of 1.76 (km/h)(2).