An Enhanced Particle Filter Algorithm with Map Information for Indoor Positioning System.

Xiaoqian Du,Xuewen Liao,Zhenzhen Gao,Ye Fan

DOI: https://doi.org/10.1109/globecom38437.2019.9013292

2019-01-01

Abstract:Recently, the demand for indoor positioning has gradually increased. Considering that people walk indoors with a serious restriction, the map information is extremely significant, which can be used as an aid in indoor positioning. In order to exploit map information thoroughly and automatically, and obtain a high-precision positioning result, we propose a map-aided particle filter (PF) algorithm based on WiFi and Pedestrian Dead Reckoning (PDR) in this paper, which exploits WiFi RSS fingerprint, inertial sensors and indoor map information comprehensively. Before the online localization, some specific image processing methods which are Morphological operation, Skeleton extraction and Line detection, are introduced to extract the latent information of indoor map, such as the skeleton of passageway in buildings, the possible forwarding directions, etc. Using the extracted features of the floor plan, the particle filter can adjust the estimated heading direction from the PDR module based on the areas of particle distribution. The real scenario experiments reveal the validity of candidate direction matching. The results also indicate that the proposed algorithm can remarkably alleviate the cumulative error, and effectively solve the problem of trajectory drift and particle deactivation during indoor positioning. Thus, compared with traditional schemes, our proposed algorithm can improve the accuracy, stability and robustness of the indoor positioning system.