Personal navigation using a kinetic model of human gait.

Abstract

Personal navigation concerns the tracking of human beings via devices carried or worn by individuals and presents a unique set of challenges in regards to navigation system and algorithm design. Many conventional position ¯xing and dead reckoning approaches tend to perform poorly given the requirements for personal navigation, which may consider GNSS-denied environments, a wide, highly dynamic range of motion, and low-cost and small form-factor sensor limitations. A novel approach to assisting or augmenting other navigation algorithms by employing a kinetic model of human gait is presented in this thesis. The kinetic model in concert with a single-axis accelerometer is shown here to comprise a virtual sensor capable of providing step size estimates in-situ for straight forward walking. Furthermore, the combination of the kinetic model and accelerometer yields a navigation solution of comparable or better performance when compared to a step counting dead reckoning approach. The derivation of this model is discussed, details of the experiments are given, and results are shown.