School of Electronic and Information Engineering, Beihang University, Beijing 100191, China.
Earth Observation System and Data Center, China National Space Administration, Beijing 100101, China.
Sensors (Basel). 2019 Jan 17;19(2):364. doi: 10.3390/s19020364.
The pedestrian navigation system (PNS) based on inertial navigation system-extended Kalman filter-zero velocity update (INS-EKF-ZUPT or IEZ) is widely used in complex environments without external infrastructure owing to its characteristics of autonomy and continuity. IEZ, however, suffers from performance degradation caused by the dynamic change of process noise statistics and heading estimation errors. The main goal of this study is to effectively improve the accuracy and robustness of pedestrian localization based on the integration of the low-cost foot-mounted microelectromechanical system inertial measurement unit (MEMS-IMU) and ultrasonic sensor. The proposed solution has two main components: (1) the fuzzy inference system (FIS) is exploited to generate the adaptive factor for extended Kalman filter (EKF) after addressing the mismatch between statistical sample covariance of innovation and the theoretical one, and the fuzzy adaptive EKF (FAEKF) based on the MEMS-IMU/ultrasonic sensor for pedestrians was proposed. Accordingly, the adaptive factor is applied to correct process noise covariance that accurately reflects previous state estimations. (2) A straight motion heading update (SMHU) algorithm is developed to detect whether a straight walk happens and to revise errors in heading if the ultrasonic sensor detects the distance between the foot and reflection point of the wall. The experimental results show that horizontal positioning error is less than 2% of the total travelled distance (TTD) in different environments, which is the same order of positioning error compared with other works using high-end MEMS-IMU. It is concluded that the proposed approach can achieve high performance for PNS in terms of accuracy and robustness.
基于惯性导航系统-扩展卡尔曼滤波-零速度更新（INS-EKF-ZUPT 或 IEZ）的行人导航系统（PNS）由于其自主性和连续性的特点，在没有外部基础设施的复杂环境中得到了广泛的应用。然而，IEZ 由于过程噪声统计和航向估计误差的动态变化而导致性能下降。本研究的主要目的是有效提高基于低成本足部安装微机电系统惯性测量单元（MEMS-IMU）和超声传感器集成的行人定位的准确性和鲁棒性。提出的解决方案有两个主要组成部分：（1）利用模糊推理系统（FIS）在解决创新统计样本协方差与理论协方差之间不匹配的问题后，为扩展卡尔曼滤波器（EKF）生成自适应因子，提出了基于 MEMS-IMU/超声传感器的行人模糊自适应 EKF（FAEKF）。相应地，自适应因子被应用于修正过程噪声协方差，以准确反映先前的状态估计。（2）开发了一种直线运动航向更新（SMHU）算法，用于检测是否发生直线行走，并在超声传感器检测到脚与墙壁反射点之间的距离时修正航向误差。实验结果表明，在不同环境下，水平定位误差小于总行程（TTD）的 2%，与使用高端 MEMS-IMU 的其他工作相比，定位误差处于同一数量级。结论是，所提出的方法可以在准确性和鲁棒性方面为 PNS 实现高性能。
