학술논문
基于伪自适应阈值零速检测法的室内个人导航系统 / Indoor Personal Navigation System Based on Zero Velocity Detection of Pseudo Adaptive Threshold
Document Type
Academic Journal
Author
Source
北京工业大学学报 / Journal of Beijing University of Technology. (9):1308-1313
Subject
Language
Chinese
ISSN
0254-0037
Abstract
为了限制惯导误差的增长，根据人行走的规律提出了一种基于个人的多种步态模式下的伪自适应阈值零速检测法.通过加速度方差和陀螺模值的平方和作为检测依据，设置一个滑动的时间窗口来判断行人的运动模式，以此来自调节阈值的大小，对跑、走、台阶等运动模式都具有稳定可靠的检测效果.由于航向角不可观测，无法使用卡尔曼零速修正算法对航向角进行修正.采用融合室内地图的粒子滤波算法使穿越墙壁的粒子权重为零，对于大误差数据，通过自适应重采样的方法来有效避免粒子的严重退化现象.实验结果表明：速度在3m/s内零速检测的准确率达到97%以上.采用融合室内地图的粒子滤波算法后，正常行走约210 m的距离室内定位误差从原来的2.3 m降低到0.2 m.
To limit error growth of the indoor navigation system ( INS ) , according to the walking regulars, based on personal multi-gait modes, a zero velocity detection of pseudo-adaptive threshold value was studied. Through the acceleration variance and the sum of the squares of the gyro modulus value as a test basis, a sliding time window was set to determine the movement patterns for adjusting the size of the threshold, and the testing results was stable and reliable for run, walk and step. Because the course angle is unobserved, the Kalman zero velocity correction algorithm cannot modify the course angle. Using the particle filter algorithm of fusing indoor maps, the particle weight across the wall is zero. For larger error data, the particle degeneration phenomenon can be effectively avoided by using the method of adaptive resampling. The experimental results show that when the speed is within 3 m/s, the zero velocity detection accuracy attains more than 97%. After using the particle filter algorithm of fusing indoor maps, about 210 m distance to walk normally, the indoor positioning error decreases from the original 2.3 m down to 0.2 m.
To limit error growth of the indoor navigation system ( INS ) , according to the walking regulars, based on personal multi-gait modes, a zero velocity detection of pseudo-adaptive threshold value was studied. Through the acceleration variance and the sum of the squares of the gyro modulus value as a test basis, a sliding time window was set to determine the movement patterns for adjusting the size of the threshold, and the testing results was stable and reliable for run, walk and step. Because the course angle is unobserved, the Kalman zero velocity correction algorithm cannot modify the course angle. Using the particle filter algorithm of fusing indoor maps, the particle weight across the wall is zero. For larger error data, the particle degeneration phenomenon can be effectively avoided by using the method of adaptive resampling. The experimental results show that when the speed is within 3 m/s, the zero velocity detection accuracy attains more than 97%. After using the particle filter algorithm of fusing indoor maps, about 210 m distance to walk normally, the indoor positioning error decreases from the original 2.3 m down to 0.2 m.