基于MEMS行人惯性导航的零速度修正技术研究

发布时间：2018-04-23 01:18

本文选题：惯性导航 + 零速度修正　；参考：《厦门大学》2014年硕士论文

【摘要】：随着科技的发展及城市化的快速推进,高林密布的大楼和大而密封的室内环境越发增多。在传统的GPS导航技术中,由于其信号在穿透建筑物后会被严重削弱,所以很难应用于室内导航与定位。然而,微机电系统(MEMS)的快速发展为这一问题带来了解决方案,基于MEMS的惯性器件以其自身体积小、成本低、功耗低等特点在室内捷联式惯性导航应用方面有了广泛研究。但是,由于MEMS惯性器件本身存在漂移、噪声等误差,所以在将其应用于室内导航的过程时,如何消除误差是较为重要的难题。本文将基于鞋绑式的捷联式惯性导航系统,拟研究行人在不同步态(正常步行、跑步)下,针对惯性器件自身的漂移误差和运动过程中引入的噪声误差进行零速度修正研究。其原理是当行人以不同步态活动时,脚步与地面接触过程中会存在一段理论上速度为零的时刻,而实际上并非为零,因此,通过检测到脚步处于零速度时刻,就可预测到误差值,然后进行剔除。零速度时刻检测部分将分别依据惯性传感器数据(加速度值与角速度值)和固定在正脚背上的超声波模块,测量其与地面距离的数据。其中依据惯性传感器数据的零速度检测算法,采用一种基于加速度平方和、加速度平方和的方差和角速度值的多条件方法,而依据超声波数据的零速度检测算法是根据对双足运动步态模型的分析,推出超声波数据模型,再根据模型总结出检测算法,从而找到零速度时刻。考虑到超声数据的稳定性,使用局部加权回归散点平滑法处理超声波数据,使其整体能够更加具体的显示出运动模型的规律和趋势,从而方便检测。通过两种零速度检测方法检测到零速度时刻后,再触发卡尔曼滤波进行误差预测并更新速度、位置和姿态信息。通过实验,其结果验证了零速度检测算法的可行性,以及在不同步态下,基于惯性数据和基于超声波数据进行零速度修正后得到的步数、单步步长、整体距离与实际情况相符。在步数方面,两者均能100%检测到；而单步步长与实际设定值的误差也不大；在整体运动距离方面,两者修正后在正常步行步态下得到了1%以内的距离误差,而跑步则达到了2%以内的距离误差。但其中,基于超声波的零速度修正得到的结果要好于基于惯性传感器数据的结果。本文对于零速度修正技术中零速度检测算法进行了不同方面的分析并验证,为MEMS惯性传感器应用到行人室内导航实际场景中提供了一种简单而高效的方法。
[Abstract]:With the development of science and technology and the rapid development of urbanization, there are more and more high-forest buildings and large and sealed indoor environment. In the traditional GPS navigation technology, it is difficult to be used in indoor navigation and positioning because its signal will be seriously weakened after penetrating the building. However, the rapid development of MEMS brings a solution to this problem. The inertial devices based on MEMS have been widely studied in the field of indoor strapdown inertial navigation because of their small size, low cost and low power consumption. However, there are drift and noise errors in MEMS inertial devices, so how to eliminate the errors is a more important problem in the process of indoor navigation. In this paper, a strapdown inertial navigation system based on shoe binding is proposed to study the zero velocity correction of drift error and noise error of inertial devices under different gait (normal walking, running). The principle is that when a pedestrian moves at a different gait, there is a time when the pace is in contact with the ground at a time when the velocity is theoretically zero, but not zero in practice, so by detecting that the pace is at zero, The error can be predicted and then eliminated. The zero-velocity detection part will measure the distance between the inertial sensor and the ground according to the data of the inertial sensor (acceleration and angular velocity) and the ultrasonic module fixed on the back of the positive foot. According to the zero velocity detection algorithm of inertial sensor data, a multi-condition method based on variance and angular velocity value of acceleration square sum, acceleration square sum is adopted. The zero-velocity detection algorithm based on ultrasonic data is based on the analysis of bipedal gait model, and then summarizes the detection algorithm according to the model to find the zero velocity time. Considering the stability of ultrasonic data, the local weighted regression scatter point smoothing method is used to process the ultrasonic data, so that the whole ultrasonic data can show the law and trend of the motion model more concretely, so it is convenient to detect. After the zero velocity time is detected by two zero velocity detection methods, the Kalman filter is triggered to predict the error and update the velocity, position and attitude information. The experimental results verify the feasibility of the zero velocity detection algorithm and the steps of zero velocity correction based on inertial data and ultrasonic data under different gait conditions. The step length is long and the whole distance is consistent with the actual situation. In terms of the number of steps, both can be detected in 100%, and the error between the single step length and the actual set value is small. In terms of the overall motion distance, the error of distance within 1% of the normal walking gait is obtained after correction. Running, on the other hand, reaches a distance error of less than 2%. But the result of zero velocity correction based on ultrasonic wave is better than that based on inertial sensor data. This paper analyzes and verifies the zero velocity detection algorithm in the zero speed correction technology, which provides a simple and efficient method for the application of MEMS inertial sensor in the pedestrian indoor navigation scene.
【学位授予单位】：厦门大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN966

【参考文献】