以行人惯导系统的姿态、速度、位移等导航误差作为研究对象,运用相关数学理论、惯性导航基本原理,综合行人运动的特点及信息融合方法,推导出导航误差之间的关联模型,建立Kalman滤波器实现误差的最优估计,通过误差校正减少累计误差,提高定位算法的精度.然后分析了比力模值与滑动方差的静止检测算法所存在的问题,设计以角速度模值、外部加速度模值为检测条件的静止检测算法,实现对步态周期静止区间的有效判断.最后设计平面定位实验和三维定位实验,验证导航算法的有效性.%In this paper we focus on the error of navigation parameters (attitude, velocity, shift, etc.) in pedestrian positioning system. Using gait characteristics and sensor fusion methods , we derive the relation model among navigation errors. Based on the relation model , we establish a novel Kalman Filter to optimal estimate navigation errors. Then we can reduce cumulative error and improve positioning accuracy by error correction. In next part we analysis the problem of SFM/SV (specific force magnitude and slip variance) static detection method. Combining angular rate magnitude and external acceleration magnitude, we design a new static detection method, which proves to work well in detecting the stance phase between steps. In the final part we conduct plane positioning experiment and three-dimensional positioning experiment. The new positioning algorithm proved to be effective and efficient.
展开▼
