Your Smartphone Could Soon Track You More Accurately Than GPS

This work introduces ReLoc-PDR, a robust pedestrian positioning platform that combines factor graph optimization, deep learning-driven visual relocalization, and dead reckoning based on inertial sensors. Through the use of smartphone sensors, the system makes step-based position estimations and periodically corrects for drift using visual observations in comparison to a pre-made 3D feature map. Even in visually impaired situations, like dim lighting or settings without texture, a fusion based on a Tukey kernel guarantees dependability. ReLoc-PDR provides precise, reliable, and real-time pedestrian trajectory estimate in a variety of indoor and outdoor situations by fusing adaptive visual updates with continuous inertial tracking.