Extracting Spatio-Temporal Information from Inertial Body Sensor Networks for Gait Speed Estimation

Abstract

The fidelity of many inertial Body Sensor Network (BSN) applications depends on accurate spatio-temporal information retrieved from body-worn devices. However, there are many challenges caused by inherent sensor errors in inertial BSNs and the uncertainty of dynamic human motion in various situations, such as integration drift and mounting error. Spatial information is especially difficult to extract from inertial data. This paper presents practical methods to minimize errors caused by these challenges within the context of a case study -- gait speed estimation -- where both temporal and spatial information are crucial for accuracy. These methods include a practical calibration procedure for correcting mounting error in order to obtain more accurate spatial information and a refined human gait model for more accurate temporal information.