行人低成本惯导自适应零速探测

Adaptive threshold zero-velocity detection for pedestrian navigation with foot-mounted lowcost inertial sensors

为了进一步提升全球卫星导航系统的多传感器融合定位精度,提出1种自适应阈值的零速探测方法:利用安装在行人足部的低成本惯性测量单元进行行人航位推算,并利用行人脚落地时的零速信息进行零速更新,以抑制误差发散;利用加速度计和陀螺仪输出构建极大似然检验统计量,发现滑动窗口内检验统计量的幅值和最小值随着速度加快而增大;然后通过幅值和最小值2个特征量,建立自适应阈值模型,阈值根据速度变化而动态调整,确保每一步均能探测零速状态,同时不会过多地探测零速;最后通过用自适应阈值和固定阈值解算不同速度的数据加以验证。结果表明:较小的固定阈值在行人速度较大时,存在大量零速间隔的漏判;较大的固定阈值则出现较多将动态判断为静态的情况;用自适应阈值可准确地探测出100%的零速间隔,明显优于固定阈值。

In order to further improve the positioning accuracy of multi-sensor fusion in global navigation satellite system,the paper proposed an adaptive threshold zero-velocity detection method:the low cost Inertial Measurement Unit(IMU)mounted on the foot of the pedestrian was used for dead reckoning,and the zero speed update was carried out to suppress error divergence by using the zero speed information of pedestrian landing;the maximum likelihood test statistics were constructed by the outputs of accelerometer and gyroscope,and it was found that the magnitude and minimum values of the test statistics in the sliding window became larger with larger speed;then the adaptive threshold model was established with the magnitude and minimum values,by which the calculated threshold can be dynamically adjusted according to the speed change,ensuring that zero velocity state can be detected in each step without too much detection as zero-velocity state for the dynamic state;finally the feasibility of the method was verified by using adaptive thresholds and fixed thresholds to solve the different speed data.Result showed that:the smaller fixed thresholds would miss many zero-velocity intervals of faster walking data,and the larger fixed thresholds would have obvious misjudgment of dynamic as static;while the adaptive threshold zero-velocity detection method would accurately detect all zero-speed intervals,which means the proposed method could be significantly better than the fixed threshold method.