多维信息感知地标匹配的PDR定位算法OA北大核心CSTPCD

Targeting at the cumulative error problem of pedestrian dead reckoning(PDR)indoor positioning system,a PDR positioning algorithm based multi-dimensional information perception landmark matching(MIPLM)is proposed.Based on the crowdsourced tracks of pedestrians in the indoor corridor environment and the prominent intersection structure,the multi-dimensional information perception landmark database is constructed from the aspects of location,heading,influence area and WiFi feature fingerprint.The proposed adap-tive landmark detection algorithm,combined with the trajectory similarity model of heading constraint,updates the pedestrian position and heading,which avoids the strong dependence on spatial position in the localization matching process.The experimental results show that,compared with other landmark construction and matching algorithms,the proposed algorithm can better reflect the correlation between pedestrian activity and indoor spatial structure.Moreover,when the initial location is unknown,the proposed algorithm can rapidly converge and provide higher positioning accuracy,which has a higher application value for the continuous positioning of indoor pedestrians.