An Energy Efficient Pedestrian Heading Estimation Algorithm using Smartphones
An Energy Efficient Pedestrian Heading Estimation Algorithm using Smartphones
复制标题
DOI：
10.1109/compsac.2019.00102
复制
发表时间：
2019-07
期刊：
影响因子：
--
通讯作者：
Yankan Yang;Baoqi Huang;Runze Yang
中科院分区：
文献类型：
--
作者：
Yankan Yang;Baoqi Huang;Runze Yang
如今，几乎每部智能手机都配备了各种惯性传感器，例如加速度计、陀螺仪、磁力计等等，这使得在智能手机上实现行人航位推算（PDR）以辅助行人定位和跟踪成为可能。然而，为了提高PDR中涉及的航向估计精度，通常会融合惯性传感器的测量值，这导致智能手机的能耗不可避免地增加，从而削弱了智能手机的续航能力。在本文中，我们提出了一种基于卡尔曼滤波器的适用于商用现成智能手机的节能型航向估计算法。具体而言，该算法利用陀螺仪在短时间内相对准确的测量值以及磁力计在长时间内相对稳定的测量值，然后将磁力计和加速度计的采样频率降低到陀螺仪采样频率的六分之一，最后基于卡尔曼滤波器融合异步惯性测量值以生成航向估计值，从而在不显著牺牲航向估计精度的情况下节省能耗。我们进行了大量实验，结果表明，与基于标准卡尔曼滤波器的算法相比，我们提出的算法平均可将能耗降低多达49.52%，同时实现了相似的航向估计精度。
Nowadays, almost every smartphone is equipped with various inertial sensors, such as accelerometer, gyroscope, magnetometer and so on, which makes it possible to implement Pedestrian Dead Reckoning (PDR) on smartphones to assist in pedestrian positioning and tracking. However, in order to improve the accuracy of heading estimation involved in PDR, it is common to fuse the measurements of inertial sensors, which results in an inevitable increase in energy consumption of smartphones, thus weakening the endurance capacity of smartphones. In this paper, we present an energy efficient heading estimation algorithm based on Kalman filter with commercial off-the-shelf smartphones. To be specific, this algorithm makes use of the relatively accurate measurements of the gyroscope in the short time and the relatively stable measurements of the magnetometer in the long term, then reduces the sampling frequency of magnetometer and accelerometer to the one-sixth of that of gyroscope, finally fuses the asynchronous inertial measurements based on Kalman filter to produce heading estimate so as to save energy consumption without significantly sacrificing heading estimation accuracy. Extensive experiments are conducted and show that our proposed algorithm reduces the energy consumption by as much as 49.52% on average compared to the standard Kalman filter based algorithm, whereas achieving the similar accuracy of heading estimation.