PDR Algorithm for Particle Filtering Based on Motion Optimization Models

In this paper, for the problem of decreased positioning accuracy of Global Navigation Satellite System (GNSS) in urban environment due to signal occlusion, reflection and other factors, combined with the demand of ordinary pedestrians for high-precision positioning, we propose a Pedestrian Dead Reckoning (PDR) combination positioning technique based on GNSS assistance. The double threshold energy detection method is used to detect the MEMS data after wavelet noise reduction, to realize accurate and efficient pace recognition, so as to improve the PDR positioning accuracy, to establish the observation model with the GNSS information, and to correct the PDR algorithm through the improved particle filtering, which can effectively inhibit the cumulative error caused by the long time working of MEMS and the positioning error caused by the reduced accuracy of GNSS in the complex environment. Experiments are designed to demonstrate the algorithm, and the pace recognition rate reaches more than 99% and the mean square error of the positioning result is 1.2m when tested in the complex test section, indicating that the algorithm can effectively improve the accuracy of positioning.