基于惯性传感器的行人室内轨迹推算与定位算法研究

发布时间：2018-11-23 11:43

【摘要】：近年来,随着基于位置服务(Location Based Service,LBS)应用的飞速发展以及微电子机械系统(Micro-Electro-Mechanical System,MEMS)技术的不断完善成熟,基于惯性传感器的行人惯性导航与轨迹追踪技术因其适用环境广、抗外界干扰能力强以及对用户无妨碍等优点,受到越来越多研究者的关注与青睐。然而,由于连续二次积分运算引起的累计误差问题严重约束了惯性导航算法的定位精度,修正推算过程中的累计误差成为了提高算法性能的关键。此外,随着智能终端硬件的更新升级,其内置的嵌入式微传感单元为其作为惯性导航与轨迹追踪算法的实现平台提供了可能。如今,以新兴智能终端为平台的行人室内轨迹追踪与定位算法逐渐成为热门研究领域。本文以惯性传感器为主要研究对象,分别针对可穿戴传感节点与智能终端两种平台中行人惯性导航与轨迹追踪算法实现改良与创新。主要研究成果如下:(1)在可穿戴传感节点平台上,针对传统捷联式惯性导航系统由于连续二次积分引起的累计误差问题,本文提出了一种以行人步伐为周期的分段式轨迹推算算法。该算法首先对行人步伐状态作划分并判别,而后在每一个步伐周期中不同状态内分别计算行人脚体的旋转角度、水平位移与偏航角,最终逐步更新行人的位置坐标并复原其运动轨迹。实际验证表明,本文算法的轨迹推算结果中终点与总距离估计误差均值分别为0.74m与1.41m,对比传统捷联式惯性导航算法在精确度方面有显著提高。(2)在智能终端平台上,提出了一种基于运动状态判别的多楼层室内环境中行人航位推算算法。该算法主要包括两个部分:第一,利用智能终端采集的加速度数据模值的极值差与无线接入点信号强度值数据来判别行人在运动过程中的不同状态。实际验证表明,本文算法运动状态误判率均值为3.74%,相比传统基于均值与方差特征的判别算法在准确率方面有显著提高;第二,根据运动状态判别结果,针对传统行人航位推算算法中步伐数目检测、步伐长度估算与航向角估算等步骤作优化与调整,实现了其基于行人运动状态的调整以及从二维平面向三维空间的扩展。实际验证表明,本文算法轨迹推算结果中坐标误差小于1.5m的步伐点比例高于85%以上,精确度较高。(3)针对本文提出的行人轨迹推算算法,分别搭建了以Shimmer传感节点为平台的行人惯性导航系统与以安卓智能手机为平台的行人多楼层室内定位系统作验证。系统实现过程主要由可穿戴传感节点校准、传感节点间数据通信、智能手机数据采集发送、地图页面定位显示等步骤组成。实际验证结果表明系统在精确度与稳定性方面均有优越表现。
[Abstract]:In recent years, with the rapid development of location-based service (Location Based Service,LBS) applications and the continuous improvement of Micro-Electro-Mechanical System,MEMS technology, Pedestrian inertial navigation and track tracking technology based on inertial sensor has been paid more and more attention and favor by more and more researchers because of its wide application environment strong ability of resisting external interference and no hindrance to users. However, the cumulative error caused by the continuous quadratic integral operation seriously restricts the positioning accuracy of the inertial navigation algorithm, and the correction of the cumulative error in the calculation process becomes the key to improve the performance of the algorithm. In addition, with the updating and upgrading of intelligent terminal hardware, its embedded micro-sensor unit makes it possible to implement inertial navigation and trajectory tracking algorithm. Nowadays, pedestrian indoor track tracking and location algorithm based on new intelligent terminal has become a hot research field. In this paper, the inertial sensor as the main research object, the wearable sensor node and intelligent terminal in the two platforms of pedestrian inertial navigation and track tracking algorithm to achieve improvement and innovation. The main research results are as follows: (1) on the wearable sensor node platform, the accumulative errors caused by the continuous quadratic integration of the traditional strapdown inertial navigation system are discussed. In this paper, a segmented trajectory estimation algorithm based on pedestrian steps is proposed. The algorithm first divides and discriminates the pedestrian step status, and then calculates the rotation angle, horizontal displacement and yaw angle of pedestrian foot body in different states in each step cycle. Finally, the position coordinates of pedestrians are updated step by step and their motion tracks are restored. The experimental results show that the mean values of the estimation errors of the end point and the total distance are 0.74m and 1.41mrespectively. Compared with the traditional strapdown inertial navigation algorithm, the accuracy of the algorithm is significantly improved. (2) on the platform of intelligent terminal, a new algorithm based on motion state discrimination is proposed to calculate the pedestrian position in multi-floor indoor environment. The algorithm mainly consists of two parts: first, using the extreme value difference of the acceleration data collected by the intelligent terminal and the signal intensity data of the wireless access point to distinguish the different states of the pedestrian in the process of motion. The actual verification shows that the average error rate of motion state of this algorithm is 3.74, which is significantly higher than the traditional discriminant algorithm based on the feature of mean and variance. Secondly, according to the result of motion state discrimination, the steps such as step number detection, step length estimation and heading angle estimation are optimized and adjusted in the traditional footpath reckoning algorithm. It is based on the adjustment of pedestrian motion state and the expansion from two-dimensional plane to three-dimensional space. The actual verification shows that the proportion of step points whose coordinate error is less than 1.5 m is more than 85%, and the accuracy is high. (3) aiming at the proposed algorithm, The pedestrian inertial navigation system based on Shimmer sensor node and the multi-floor indoor positioning system based on Android smart phone are built. The system is mainly composed of wearable sensor node calibration, sensor node data communication, smart phone data acquisition and transmission, map page location and display, and so on. The experimental results show that the system is superior in accuracy and stability.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN96;TP212

【参考文献】