基于伪标准差和N-P准则的行人导航零速检测

Pedestrian navigation zero-velocity detection based on pseudo-standard deviation and N-P criterion

针对行人导航零速检测存在误判、漏判且判别阈值难以随行走速度及不同行人在线调整的问题,提出了一种基于伪标准差和N-P(Neyman-Pearson)准则的检测方法。首先通过分析行走过程中陀螺仪模值的特性,构造了一个融合了两个特征量的伪标准差,提高了零速区间与非零速区间的区分度,降低了误判和漏判的概率。然后通过行走实验数据分别得到了零速区间和非零速区间的伪标准差概率密度曲线,将零速检测转化为一个二分类问题,利用N-P准则推导得到这个二分类问题的判别规则和判别阈值。最后经过不同行人、不同速度的行走实验,验证了该方法准确有效,并能适应多种步速,平均检测精度达到了99.67%,与传统的三条件、四条件检测法相比,降低了零速检测算法的复杂性、提高了检测精度,而且增强了算法在不同行走速度和不同行人间的适应性。

In view that the zero-speed detection of pedestrian navigation has the problems of misjudgment and omission, and the discriminant threshold is difficult to be online adjusted according to different walking speeds and pedestrians, a novel detection method based on pseudo-standard deviation and N-P(Neyman-Pearson) criterion is proposed. Firstly, by analyzing the characteristics of the gyroscope mode value during walking, a pseudo standard deviation combining two characteristic quantities is constructed, which improves the discrimination between the zero-velocity interval and the non-zero-velocity interval, and reduces the probability of misjudgments and missed judgments. Secondly, the pseudo-standard probability density curves of the zero-velocity interval and the non-zero-velocity interval are obtained respectively from walking experimental data. The zero-velocity detection is converted into a binary classification problem. The discriminant rules and the discriminant threshold of the binary classification problem were derived by using the N-P criterion. Finally, the walking experiments of different pedestrians and speeds verify that the proposed method is accurate and effective, and can adapt to a variety of walking speeds. The average detection accuracy reaches 99.67%. Compared with the traditional three-condition and four-condition detection methods, the complexity of the zero-velocity detection algorithm is reduced, the detection accuracy is improved, and the adaptability of the algorithm to different walking speeds and different pedestrians is strengthened.