基于MIMU的适应多步态的步行导航算法

针对目前国内外步态导航算法中引入的零速检测大都不能很好的对多种步态的行走进行辨识的问题，提出了基于低成本MIMU（微惯性测量单元）、且能兼容多种步态的步行导航算法。算法通过采集安置在行人脚部MIMU输出的测量信息，用捷联惯性积分算法进行导航解算。其间，提出一种新的零速检测方法对行走时变化的步速和步型进行准确辨识，进而找到脚步的零速时刻点，并通过设计的扩展卡尔曼滤波（EKF）对导航解算结果进行零速修正（ZUPT），实现系统的反馈。最后进行两组实验对算法验证。结果表明，该步态导航算法能对行走时的多步态问题有很好的兼容性，零速修正时刻辨识准确度高，两组实验的导航解算误差均达到0.6%以内，进一步提高了步态导航算法的精度和实用性。

A Multi-gait Adapting Pedestrian Navigation Algorithm Based on MIMU

In view of the problem that the majority stance phase detectors in previous pedestrian navigation algorithms commonly have a weak precision in walking with different gaits, a novel pedestrian navigation algorithm based on a low-cost MIMU (Micro Inertial Measurement Unit) was proposed, which is compatible with multi-gait. Through obtaining original measurement data from foot-mounted MIMU, pedestrian navigation information is calculated depending on strapdown inertial navigation algorithm. During the movement of people, a new stance phase detectors was carried out, which can provide a better precision of zero velocity measurement when people holding variable speed or steps. And then relying on a designed Extended Kalman Filter (EKF) implements zero velocity update (ZUPT) to correct navigation result and to build system feedback. In the end, the algorithm has been verified by two experiments. The results indicate that the novel algorithm is effective to solve the problem of multi-gait in pedestrian navigation, and the positioning errors of the experiments are less than 0.6%. Besides, the algorithm improves the precision and practicability of applications.