GNSS/里程计组合定位抗差估计法

利用Kalman滤波进行GNSS/里程计组合定位难以实现两种传感器的最优利用。针对轮轨相对运动时里程计数误差增大的情况,提出了对GNSS位置导出里程、速度导出里程和里程计输出里程使用抗差最小二乘求取最优里程;并使用中位数求取抗差迭代初值。增加GNSS测速里程观测值,提高了观测冗余,有利于计算抗差初值。抗差等价权能够平衡3种观测值的比重,增强鲁棒性。计算结果表明,引入速度信息的抗差解法优于单一传感器定位解;也优于传统滤波解法。

GNSS/Odometer Integrated Positioning Using Robust Estimation

GNSS/Odometer is often integrated through Kalman filter but this integration is not optimal. Since relative motions between the rail and wheel produce errors exhibiting the property of outliers, we proposed solving the optimal distance by robust least squares estimation using distance information from GNSS position, GNSS veloc-ity and odometer and the initial value is determined by the median function. The additional velocity-derived dis-tance improve the redundancy and serves well for calculation the initial value of robust estimation process. The e-quivalent weight of the robust estimation balances the three different measurements well and enhances the robust-ness of integrated system. The results show that the proposed robust solution aided by velocity outperforms the posi-tioning accuracy of either of the two sensors and that of the traditional filtering solution.