Overview of methods for visual-aided pedestrian navigation

Pedestrian navigation is often needed in surroundings where GPS signals are degraded or not available at all, like indoors and urban canyons. Aiding from other location sensors is then needed. Self-contained sensors with pedestrian dead reckoning algorithms offer location information that doesn't depend on any infrastructure and isn't restricted to certain areas. The problem with those sensors is, however, that they suffer from drifting in time and are prone to errors. Integrating an independent drift-free sensor like a camera and correcting the heading with information derived from it will provide a seamless navigation system. In this paper we will give an overview of the systems developed for pedestrian visual-aided navigation for GPS degraded or denied environments. We will also introduce methods and algorithms for retrieving position and motion information from images as well as for integrating this information with measurements obtained from other location sensors.