DREAR - Towards Infrastructure-Free Indoor Localization via Dead-Reckoning Enhanced with Activity Recognition

The recent indoor localization techniques use inertial sensors for position estimations in order to obtain a certain degree of freedom from infrastructure based solutions. Unfortunately, this dependency cannot be completely eliminated due to the cumulative errors induced in the localization process. While many methods are designed to reduce the required number of reference points or try to offer unsupervised maintenance, completely infrastructure independent solutions are still missing. In this paper we propose a novel approach for mobile-based indoor localization and navigation services by introducing a context-aware localization framework. We exploit the ability to recognize certain human motion patterns with a smartphone, representing activities related to walking, climbing stairs, taking escalators, etc. This allows the detection of corridors, staircases and escalators, knowledge which can be used to create building interior related reference points. Based on these a scenario specific context interpreter controls the localization process and provides position refinement for the elimination of the cumulated errors. Using our solution the auxiliary reference points can be omitted, thus, a completely infrastructure-free localization system is formed. The proposed solution is evaluated in a subway scenario and its performance is analysed focusing on the influence of dead-reckoning error on path reconstruction, the effect of activity detection quality on localization performance, respectively the benefits of using additional context-related information. The results are promising, our solution presents good localization and path reconstruction performance, showing potential for real-world scenarios, where the deployment of auxiliary localization infrastructure is unfeasible.