在室内定位系统中,基于Wi-Fi技术的定位精度很大程度上依赖于信号的稳定,信号的多径效应与非视距(Non Line of Sight,NLOS)会增大定位误差.行人航位推算(Pedestrian Dead Reckoning,PDR)定位系统会因传感器自身误差与噪声产生累计误差.针对上述问题,提出了一种改进的PDR与最小一乘法(Least Absolute Deviation,LAD)融合的室内定位算法.该算法基于模糊逻辑将PDR算法的步长固定参数改进为变量参数,同时根据LAD的定位结果对PDR进行周期性位置与拐点位置校正,选择扩展卡尔曼滤波(Extend Kalman Filter,EKF)将改进的PDR与LAD进行融合,以降低PDR的累计误差与LAD的突变误差,提高定位精度.实验结果表明:所提方法较其他方法具有更高的定位精度.
A Positioning Method for Improved PDR and Least Absolute Deviation Fusion
In an indoor positioning system,the positioning accuracy based on Wi-Fi largely depends on the stability of the signal,and the multipath effect and non line of sight(NLOS)will increase the positioning error.Pedestrian dead reckoning(PDR)positioning system will generate cumulative error due to sensor error and noise.Targeting at the above problems,an indoor localization algorithm based on the fusion of improved PDR and Least Absolute Deviation(LAD)is proposed.The fixed step parameter of the PDR algorithm is modified into a variable parameter according to the fuzzy logic,and the periodic position and inflection point position of the PDR are corrected ac-cording to the positioning result of the LAD,and the Extended Kalman Filter(EKF)is selected to fuse the improved PDR with the LAD to reduce the cumulative error of the PDR and the mutation error of the LAD.The experimental results show that the algorithm can ef-fectively improve the positioning accuracy.
indoor positioningPDRWi-Fithe Least Absolute Deviationfuzzy logic
国家科技期刊平台
NSTL
万方数据
