【摘要】 针对惯性室内定位技术无法获得绝对位置和长时间累计误差较大的缺点,提出一种将惯性定位技术与WiFi技术融合定位的方法:通过粒子滤波算法将惯性定位方法得到的相对位置结果作为观测量对位置进行估计,其中通过WiFi定位结果和磁场方向对粒子滤波进行初始化,并结合墙壁信息计算概率转移密度来辅助定位;离线阶段仅需要在特定区域建立指纹库,并根据建筑结构在走廊大厅等位置划定不可穿墙区域,不需要庞大的建库成本。实验结果表明该方法不仅可以弥补惯性定位技术中无法获得绝对位置的缺点,而且能够减少建立大范围WiFi指纹库的人力成本,提高整体定位精度。更多还原

【Abstract】 Aiming at the problem that inertial measurement unit(IMU)based indoor positioning method cannot provide enough accuracy because of the accumulative error,the paper developed an integrated positioning system utilizing WiFi and MEMS-IMU:the positioning results of WiFi and MEMS-IMU were used as observed factors in the particle filter algorithm,among which the particles were initialed with the magnetic field information and the positioning results of WiFi,and the probability transition density was calculated with the wall information for assisting the positioning process.Furthermore,the fingerprint database would be needed only in some important areas which reduces the manpower of offline phase.Experimental result showed that the method could not only improve the accuracy of indoor positioning but also decrease the manpower of building WiFi fingerprint database.更多还原