基于级联滤波的行人惯性室内定位方法

A pedestrian inertial indoor location method based on cascade filters

针对目前足部安装式行人惯性导航系统对精度的需求,通过融合惯性导航传感器和室内平面地图信息,提出一种级联结构的扩展卡尔曼滤波和非递归贝叶斯滤波的室内定位方法。首先,在下级扩展卡尔曼滤波器中,使用零速检测算法检测行人运动的零速区间,在零速区间内,使用基于扩展卡尔曼滤波的启发式航向修正、零速更新、零角速度更新三种算法,初步修正惯性导航的航向和距离误差。然后,在上级非递归贝叶斯滤波中,融合地图先验信息和下级滤波器估计结果,使用非递归贝叶斯滤波地图匹配算法对导航误差进一步校正。实验结果表明,与零速修正惯性行人定位相比,级联导航系统的航向漂移很小,平均距离误差也从12.40 m下降到了0.45 m。

To deal with the accuracy requirements of present foot-mounted pedestrian inertial navigation systems,an indoor localization method with cascaded structure of extended Kalman filter and non-recursive Bayesian filter is proposed by fusing inertial navigation sensors and indoor map information.Firstly,the zero velocity interval of pedestrian motion is detected using the zero velocity detection algorithm in the lower extended Kalman filter,and the three algorithms of heuristic heading reduction,zero velocity update,and zero angular velocity update based on the extended Kalman filter are used in the zero velocity interval to initially correct the heading and distance errors of inertial navigation.Then,the map prior information and the lower filter estimation results are fused,and the navigation error is further corrected using the non-recursive Bayesian filter map matching algorithm in the upper non-recursive Bayesian filter.The experimental results show that the heading drift of the cascaded navigation system is small compared to the Inertial pedestrian positioning based on zero velocity correction,and the average distance error decreases from 12.40 m to 0.45m.