ReLoc-PDR: A Robust Smartphone System for Indoor Pedestrian Positioning

In this research, we introduce ReLoc-PDR, a robust pedestrian positioning method that uses graph optimization to combine optical relocalization and pedestrian dead reckoning (PDR). To ensure consistent and dependable localization, ReLoc-PDR efficiently manages anomalous visual observations and reduces cumulative drift by utilizing deep learning-based descriptors and a Tukey kernel fusion process. Unlike infrastructure-dependent approaches, this smartphone-based technology enables reliable and accurate pedestrian monitoring in visually degraded and satellite-denied situations, including nighttime and texture-less hallways. ReLoc-PDR is a potential method for useful, infrastructure-free indoor navigation since real-world tests confirm the system's exceptional performance.