Wearable inertial sensors are a convenient way to track pedestrians in situations where other localisation systems fail. More and more researchers are using inertial pedestrian tracking as a component in more complex applications. But despite the many publications on the topic, and the increasing availability of inertial sensors, implementing a reasonably accurate tracker can be challenging for researchers without a background in inertial navigation and Kalman filtering. With this article, we aim to provide fellow researchers with a simple yet effective
implementation which works out of the box, but with enough clear explanations and references to allow for further investigation if required.
We discuss common mistakes and practical aspects, such as parameter tuning and choice of sensors, which we found to be lacking in other work. The Matlab implementation and a few sample recordings are available online at http://eis.comp.lancs.ac.uk/pdr/.