Automated gait segmentation and tracking using inertial measurement units

Abstract

In this thesis, a methodology is presented to automate the labelling, event detection, segmentation, tracking, and parameter extraction of IMU gait data for sensors placed on the feet and shanks. The algorithms presented were tested using IMU data from three different styles of gait, normal gait, antalgic gait, and limited mobility gait. The algorithms developed were found effective for all of the simulated gait styles without mislabelling or detecting erroneous gait segments. The resultant gait trajectories and parameters were analyzed and were found to accurately depict the differences between each of the different styles of gait.

The methodology presented can be used for the rapid and accurate processing of gait data for multiple styles of gait. This quantification of gait data can enable the collection of IMU gait data on a larger scale. This provides an accessible, low-cost option for out-of-laboratory gait data collection.