ReLoc-PDR: Visual Relocalization Enhanced Pedestrian Dead Reckoning via Graph Optimization

Abstract-- Accurately and reliably positioning pedestrians in satellite-denied conditions remains a significant challenge. A graph optimization-based fusion mechanism with the Tukey kernel effectively corrects cumulative errors and mitigates the impact of abnormal visual observations. Real-world experiments demonstrate that our ReLoc-PDR surpasses representative methods in accuracy and robustness, achieving accurte and robust pedestrian positioning results using only a smartphone in challenging environments such as less-textured corridors and dark nighttime scenarios. The pursuit of a low-cost, robust, and self-contained positioning OBUST and accurate indoor pedestrian navigation plays a crucial role in enabling various location-based services system is of great importance to flexible and resiliant (LBS). Visual relocalization, which estimates environments is a fundamental requirement for numerous the 6 degree-of-freedom (DoF) pose of a query image against applications, including emergency rescue operations, an existing 3D map model, holds potential for achieving driftfree path guidance systems, and augmented reality experiences [1], global localization using only a camera. The existing indoor positioning solutions relying on the smartphones commonly are with built-in cameras, utilizing deployment of dedicated infrastructures are susceptible to signal image-based localization methods on mobile devices becomes interference and non-line-of-sight (NLOS) conditions, and a viable approach. However, the challenge lies in the susceptibility their widespread deployment can be prohibitively expensive. of image-based relocalization to environmental factors In contrast to infrastructure-based positioning methods, such as changes in lighting conditions and scene dynamics.