Position Tracking During Human Walking Using an Integrated Wearable Sensing System

Abstract

Fixed motion tracking systems can offer highly accurate data but several drawbacks are present, including a high upfront cost and require the user to stay within a very limited area. Of keen interest are shoe mounted systems which aim to offer a similar suite of information but are unconstrained in their operating environment. The potential of knowing the user's foot placement and orientation is an extremely valuable set of information. This data can be used in a wide range of applications such as healthcare monitoring, emergency responder localisation, and lower limb prosthetic stability and control. This thesis investigates the potential of using low cost (~£30) inertial measurement units (IMUs) to track a user's motion and position. When using an IMU, general purpose strap-down navigation is shown to give inadequate results after only seconds of use. Thus, to provide corrections, an Extended Kalman filter (EKF) is used to provide zero velocity and heuristic drift reduction updates. This system is shown to have typical loop closure errors of smaller or equal to1% with maximum errors of 4-5%. In parallel with the IMU an ultrasound (US) trilateration system calculates the displacement of each step and the results of the IMU and US systems are combined. This addition gave slight improvements in the results, typically reducing the cumulative error over a walk by 15%. Lastly, a particle filter can impose movement constraints on the predicted motion by including environmental information. In combination with the previous two sensing systems the addition of a particle filter gave consistent errors of

Date of Award	31 Dec 2016
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Lei Ren (Supervisor) & Andrew Weightman (Supervisor)