SugarTrail: Indoor navigation in retail environments without surveys and maps

A system that helps people navigate in indoor environments on a fine-grained level can enable a variety of pervasive computing applications in retail environments. Existing indoor navigation systems rely on extensive RF tagging surveys and accurate floor plans. These prerequisites are often impractical in indoor environments. In this paper, we present SugarTrail, a system for indoor navigation assistance in retail environments that minimizes the need for active tagging and does not require existing maps. By leveraging the structured movement patterns of shoppers in retail store environments, the system provides higher accuracy than existing radio finger-printing approaches. With minimal setup and active user participation, the system automatically learns user movement pathways in indoor environments from radiofrequency and magnetic signatures. These pathways are clustered and used to automatically build a navigable virtual roadmap of the environment. We present results from a campus testbed and from actual radio measurements collected in an operational supermarket to show that SugarTrail system can navigate users with a success rate of > 85% and an average accuracy of 0.7m.