基于MEMS惯性测量单元的多源信息自适应步数检测方法

Multi-source information adaptive step detection method based on MEMS inertial measurement unit

针对基于MEMS惯性测量单元的行人航迹推算中步数检测方法仅利用单一的加速度信号检测精度较低的问题,提出一种多源信息自适应步数检测方法。该方法通过综合考虑人体运动过程中的加速度信号和角速度信号,根据不同的步态特征通过设定不同的自适应阈值条件实现步数的检测。虽然常规的峰值检测算法和固定阈值检测算法在单一步态下步数检测精度相对较高,但是对复杂运动状态下的步数检测精度很差,无法适用于真实的行人运动过程中步数的检测。然而多源信息自适应步数检测方法却能够在行人不同运动状态下精确检测步数,该方法明显优于常规的峰值检测方法和阈值检测方法。试验结果表明,本文提出的多源信息自适应阈值检测方法在行人不同运动状态下的步数检测精度可达98%以上。

In pedestrian dead reckoning（PDR） based on MEMS IMU, the detection for number of steps has a lower accuracy due to basing only on the accelerometer signals. To improve the detection accuracy, an adaptive detection method with multi-source information is proposed for the step detection. In this method, the adaptive detection is realized by comprehensively considering the angular velocity signals and the acceleration signals in the process of human movement and setting different adaptive threshold conditions based on different gait features. Unlike the conventional peak detection algorithm and the fixed-threshold detection algorithm, which have rather poor detection accuracy under abnormal human movements, the proposed method can accurately detect the number of steps under complex pedestrian motion behaviors. Experiment results show that the detection precisions by the proposed method can reach more than 98% under different pedestrian motion states.