Pedestrian GraphSLAM using Smartphone-based PDR in Indoor Environments

Simultaneous Localization and Mapping (SLAM) for pedestrians is a relatively new approach for the indoor localization problem. With the advancements in smartphone technology, pedestrian SLAM has transitioned towards utilizing smartphones' integrated sensors through Pedestrian Dead-Reckoning (PDR) techniques. In this paper, we present a novel approach for indoor user localization, trajectory tracking and mapping through GraphSLAM: modeling the spatial structure of a user's positions as a graph optimization problem. The paper proposes (1) a new algorithm for calibrating the heading measurements acquired through the smartphone sensors for PDR, and (2) a heading-detection stage as a pre-processing stage for GraphSLAM. Experiments were conducted using an iPhone 7 within an academic building with different users. The proposed algorithms were able to overcome the drift errors in heading measurements and provide accurate estimates for users' locations and movement trajectories.