TIGHTLY-COUPLED INTEGRATION OF BLE AND PDR USING GRAPH OPTIMIZATION FOR INDOOR PEDESTRIAN NAVIGATION

Abstract

In this paper, we propose an indoor navigation method based on the tightly-coupled (TC) integration of Bluetooth low energy (BLE) and pedestrian dead reckoning (PDR) using a graph optimization model. We first utilize the Gaussian probability model to update the particle weights that considers the ranging model’s estimation performance at different distances to determine the particle weight. Moreover, the BLE walking-surveyed or crowdsourced landmarks, combined with accurate ranging of BLE at a short distance, is used to construct a graph optimization model, and the Levenberg-Marquardt (LM) algorithm is adopted to optimize this model to improve track tracking performance. The performance of the proposed algorithm has been verified in the hallway scene and another challenging room scene. The results show that compared with the standard particle filter (PF) method, the average positioning accuracy of the proposed algorithm is improved by 64.0% and 54.75%, and the error variance is significantly reduced by 76.23% and 68.60%, respectively, which is a significant improvement in both robustness and accuracy. Furthermore, the test shows that the proposed method can calculate reasonable trajectories even in complex room scenarios.