Optimal Robust Positioning Using Factor Graph

Abstract

Smartphone GNSS positioning accuracy is often limited by the use of low-cost antennas and chips, resulting in significant impact of signal degradations due to non-line-of-sight (NLOS) and multipath effects. Traditional techniques, such as extended Kalman filtering (EKF), often struggle to maintain accuracy, especially in urban settings. However, factor graph optimization (FGO) has emerged as a promising solution, demonstrating robust navigation capabilities in urban environments by leveraging multi-epoch GNSS measurements to estimate user positions concurrently. Therefore, we will implement an optimal robust positioning solution using FGO framework in this study. The integration of the Inertial Measurement Unit (IMU) and Global Navigation Satellite System (GNSS) measurements is achieved through an approach that employs the factor graph model. Unlike traditional filtering algorithms, the factor graph framework leverages all preceding measurements for state estimation. This study conducts experiments utilizing the Google Smartphone Decimeter Challenge 2023 dataset as its basis. This research aims to develop an optimal and resilient positioning algorithm by integrating GNSS and IMU through a loosely coupled approach leveraging FGO formulation with directly combining position data from GNSS receivers with the IMU measurements. This method of integration is easily implementable in both hardware and software. We compare this to the use of Kalman Filter/RTS for the same integration process. The Kalman Filter/RTS results gave an accuracy of 2.603 m while the accuracy of FGO implementation was 1.661 m. This research achieved an approximately 56.71% increase in accuracy with the implemented FGO approach.