基于贝叶斯网络的强鲁棒性零速检测方法

A Robust Zero Velocity Detection Method Based on Bayesian Networks

摘要:

在行人惯性导航系统中,零速检测是实现速度误差清零和导航误差估计的前提,有着重要的作用；针对现有的零速检测方法检测精度低、鲁棒性差的问题,采用有效的步态周期分割方法并且引入基于惯性传感器测量值和运动学知识的贝叶斯网络模型来推断零速区间；该方法可以有效减少零速(ZV)边界的模糊性,提高零速检测的精度,增强零速检测的鲁棒性；实验表明行人以较高的速度行走时,基于贝叶斯网络的零速检测方法零速错误检测去除率提高,零速检测效果好。

Abstract:

In the pedestrian inertial navigation system, the zero velocity detection is the premise to reset the velocity errors and estimate the navigation errors, so it plays an important role in the system. Aiming at the problems of low accuracy and poor robustness in the existing zero velocity detection method, the proposed method adopts an effective gait cycle segmentation method and introduces a Bayesian network (BN) model based on the measurements of inertial sensors and kinesiology knowledge to infer the ZV period. This method can effectively reduce the ambiguity of the zero velocity (ZV) boundaries, and enhance the robustness of the zero velocity detection. The experiments reveal that the removal rate of ZV false detections by zero velocity detection method based on BN increases and the method is effective at high walking speed.