一种基于抗差滤波的行人导航算法研究

行人导航系统(pedestrian navigation system, PNS)通常采用全球定位系统(global positioning system, GPS)和航位推算(dead reckoning, DR)组合导航的方式进行定位，因此其定位精度易受GPS定位误差特别是定位粗差的影响。为了减小这种影响，提高行人导航系统定位精度，采用了一种基于抗差滤波的GPS/DR组合行人导航算法。该算法首先对DR系统误差建模，获得行人导航系统卡尔曼滤波模型，再通过GPS与DR系统观测量之差，估计当前观测噪声与先验统计特性的符合程度，利用等价权实时调整观测权值，以避免观测粗差对组合导航精度的影响。最后通过对实测数据的分析表明，在GPS定位误差较大或含粗差情况下，该方法较卡尔曼滤波算法能明显抑制定位误差的影响，将定位精度提高5 m左右，能够在不增加硬件的基础上有效提高GPS/DR组合行人导航精度。

Method of pedestrian navigation based on robust filter

Since the positional method of pedestrian navigation system (PNS) is usually assisted by global positioning system (GPS) and dead reckoning (DR) system integrated navigation algorithm, the PNS positional accuracy tend to be diminished by the GPS positional errors, especially positional outliers. To reduce the tendency and improve the accuracy of the pedestrian navigation system, a method of GPS and DR integrated navigation algorithm based on robust filter is presented. The Kalman filter model of PNS is obtained through modeling DR systematic errors and DR sensor errors first. Then, the positional difference between GPS and DR is used to estimate the extent to which the latest observation error meets their prior statistics. Based on 〖JP2〗the extent, the observed weight is renewed by an equivalent weight to restrict the negative effect of GPS positional error on integrated navigation. Finally, the result of real data from the experimental prototype shows that, in the case of GPS performing badly, the robust method is much more effective than Kalman filter in reducing the influence of the GPS positional error and could improve the positional accuracy by about 5 m. Thus, the positional accuracy of PNS based on GPS/DR integrated navigation is improved without additional hardwares.