基于足间距信息辅助的行人三维惯性定位算法

针对行人惯性导航系统误差随时间累积致使定位精度严重下降的问题，提出了一种基于足间距信息辅助的行人三维惯性定位算法。该算法在零速修正算法的基础上，利用足部安装的超声波测距模块实时测量行人双足相对距离，构建了基于超声测距的足间距约束模型，通过随机森林算法实现行人运动模式识别，并针对上下楼梯场景，利用台阶高度和足间距信息进行高度解算，最终实现行人三维惯性定位。在实际路线上开展了三维定位实验，数据显示，所提算法平面闭环误差为总路程的0.64%，与零速修正算法相比下降了55.56%，高度误差为0.06 m，与零速修正和气压计联合算法相比下降了64.70%，能够实现导航误差在总路程的0.50%以内的三维定位。实验结果表明，所提算法具有良好的工程应用价值。

Three-dimensional pedestrian inertial positioning algorithm assisted by foot spacing information

To solve the problem of the severe decrease in positioning accuracy caused by the accumulation of errors in the pedestrian inertial navigation system over time, a three-dimensional (3D) pedestrian inertial positioning algorithm based on the auxiliary information of inter-foot distance is proposed. Based on zero velocity update (ZUPT), the algorithm utilizes an ultrasonic ranging module installed on the feet to measure the relative distance between the pedestrian''s feet in real-time, constructing a constraint model based on ultrasonic ranging for inter-foot distance. In addition, the random forest algorithm is used to recognize pedestrian motion patterns. For stairway scenarios, the height is calculated based on the step height and inter-foot distance, ultimately achieving 3D pedestrian inertial positioning. The 3D positioning experiment is carried out on the actual route. The data show that the plane closed-loop error of the proposed algorithm is 0.64% of the total distance, which is decreased by 55.56% compared with the ZUPT algorithm, and the height error is 0.06 m, which is decreased by 64.70% compared with the ZUPT combined with barometer algorithm. In general, the 3D positioning with navigation error within 0.50% of the total distance can be achieved. These results demonstrate the practical engineering value of the proposed algorithm.