基于四元数和卡尔曼滤波的姿态角估计算法研究与应用
发布时间：2018-08-31 09:15
【摘要】：姿态估计最早应用于军事航天领域,并在该领域的研究比较成熟,但由于传统惯性传感器(陀螺仪和加速度计)体积较大、成本高、携带不便等原因影响了姿态估计在其他领域的应用。近年来,随着MEMS(Micro Electro Mechanical System,微机械电子系统)技术的发展,特别是MEMS惯性传感器和MEMS磁强计的出现,扩展了姿态估计在其他领域的应用,例如在医学领域中的内窥镜姿态定位技术,虚拟与现实领域中的情景交互、手势识别等。姿态指的是一个坐标系和另外一个坐标系之间的位置关系,常用一组姿态角:俯仰角、横滚角和偏航角来描述。里面涉及到姿态角的求解,即姿态解算问题和姿态角的精度问题,本文应用姿态角检测跌倒事件的发生,就是以姿态角计算的准确性为前提。描述姿态的方法有很多,包括欧拉角法、方向余弦法和四元数法等。欧拉角描述姿态会出现奇异性问题,方向余弦法计算量较大,目前广泛采用四元数法,本文基于四元数法,利用四阶龙格库塔法求解四元数微分方程得到姿态角,实现姿态解算。由于MEMS陀螺仪自身的特性,静态情况下其输出的信号中包含常值误差和随机漂移误差,这就会造成在采用陀螺仪计算角度时会产生角度漂移,无法实现长时间的精确测量;同样,加速度计根据重力场计算出水平倾角(俯仰角和横滚角),而实际应用中载体都是运动的,势必会引入线性加速度,造成水平倾角计算的不准确,磁强计根据地球磁场计算出偏航角,但由于地磁场容易受到干扰,那么计算的偏航角也不会太准确。为解决以上问题,目前广泛采用多传感器信息融合的方式进行姿态角估计,本文在参考前人融合算法的基础上,提出了自己的融合方案:采用卡尔曼滤波算法对三轴陀螺仪、三轴加速度计和三轴磁强计进行信息融合,得到最优估计角度。首先,根据传感器输出的数据建立随机漂移AR误差模型,对原始数据进误差补偿和滤波,然后将陀螺仪计算的角度作为估计角度,将加速度计和磁强计计算的角度作为测量角度,利用卡尔曼滤波算法融合估计角度和测量角度,得到最优的姿态角估计。并对提出的融合算法进行实际验证,达到预期效果。最后,按照设计的算法,利用九轴传感器MPU-9150和Arduino Pro mini设计了可佩戴于腰部的姿态角检测装置,实现对人体跌倒姿态的检测,通过实际测试证明了其可靠性并提出了改进的意见。本文作为一篇工程应用文,将姿态角估计应用于人体跌倒姿态的检测中,为姿态估计在其他领域的应用起到了推动作用,有一定的应用价值和实际意义。
[Abstract]:Attitude estimation was first applied in the military space field, and the research in this field is mature. However, the traditional inertial sensors (gyroscopes and accelerometers) are large in size and high in cost. The inconvenience of carrying affects the application of attitude estimation in other fields. In recent years, with the development of MEMS (Micro Electro Mechanical System, technology, especially the emergence of MEMS inertial sensors and MEMS magnetometers, the application of attitude estimation in other fields, such as endoscope attitude positioning technology in medical field, has been expanded. Scene interaction, gesture recognition and so on in the field of virtual reality. Attitude refers to the position relationship between one coordinate system and another coordinate system, which is usually described by a set of attitude angles: pitch angle, roll angle and yaw angle. In this paper, attitude angle is used to detect the fall event, which is based on the accuracy of attitude angle calculation. There are many methods to describe attitude, including Euler angle method, directional cosine method and quaternion method. The singularity problem occurs in Euler angle describing attitude, and the direction cosine method has a large amount of calculation. At present, the quaternion method is widely used. Based on the quaternion method, the attitude angle is obtained by solving the quaternion differential equation with the fourth order Runge-Kutta method. The attitude calculation is realized. Because of the characteristic of MEMS gyroscope, the output signal of MEMS gyroscope contains constant error and random drift error in the static condition, which will result in angle drift when using gyroscope to calculate the angle, which can not realize the accurate measurement for a long time. In the same way, the accelerometer calculates the horizontal inclination angle (pitch angle and roll angle) according to the gravity field. However, in practical application, the carrier is moving, which will inevitably introduce linear acceleration, resulting in the inaccuracy of the calculation of the horizontal inclination angle. The yaw angle of the magnetometer is calculated according to the earth's magnetic field, but because the geomagnetic field is easily disturbed, the calculated yaw angle is not too accurate. In order to solve the above problems, multi-sensor information fusion is widely used to estimate the attitude angle. In this paper, based on the previous fusion algorithms, our own fusion scheme is proposed: the Kalman filter algorithm is applied to the three-axis gyroscope. The three-axis accelerometer and three-axis magnetometer are fused to obtain the optimal angle estimation. Firstly, the random drift AR error model is established according to the output data of the sensor, and the error compensation and filtering of the original data are made. Then, the angle of the gyroscope calculation is used as the estimation angle. The angle calculated by accelerometer and magnetometer is taken as the measuring angle, and the optimal attitude angle estimation is obtained by combining the estimation angle with the measurement angle using Kalman filter algorithm. The proposed fusion algorithm is verified to achieve the desired results. Finally, according to the designed algorithm, the attitude angle detection device which can be worn on the waist is designed by using MPU-9150 and Arduino Pro mini, which can be worn in the waist. The reliability of the device is proved by the actual test and the improved advice is put forward. In this paper, as an engineering application, attitude angle estimation is applied to the detection of human fall attitude, which plays an important role in the application of attitude estimation in other fields, and has certain application value and practical significance.
【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN713;TP212
本文编号：2214556
[Abstract]:Attitude estimation was first applied in the military space field, and the research in this field is mature. However, the traditional inertial sensors (gyroscopes and accelerometers) are large in size and high in cost. The inconvenience of carrying affects the application of attitude estimation in other fields. In recent years, with the development of MEMS (Micro Electro Mechanical System, technology, especially the emergence of MEMS inertial sensors and MEMS magnetometers, the application of attitude estimation in other fields, such as endoscope attitude positioning technology in medical field, has been expanded. Scene interaction, gesture recognition and so on in the field of virtual reality. Attitude refers to the position relationship between one coordinate system and another coordinate system, which is usually described by a set of attitude angles: pitch angle, roll angle and yaw angle. In this paper, attitude angle is used to detect the fall event, which is based on the accuracy of attitude angle calculation. There are many methods to describe attitude, including Euler angle method, directional cosine method and quaternion method. The singularity problem occurs in Euler angle describing attitude, and the direction cosine method has a large amount of calculation. At present, the quaternion method is widely used. Based on the quaternion method, the attitude angle is obtained by solving the quaternion differential equation with the fourth order Runge-Kutta method. The attitude calculation is realized. Because of the characteristic of MEMS gyroscope, the output signal of MEMS gyroscope contains constant error and random drift error in the static condition, which will result in angle drift when using gyroscope to calculate the angle, which can not realize the accurate measurement for a long time. In the same way, the accelerometer calculates the horizontal inclination angle (pitch angle and roll angle) according to the gravity field. However, in practical application, the carrier is moving, which will inevitably introduce linear acceleration, resulting in the inaccuracy of the calculation of the horizontal inclination angle. The yaw angle of the magnetometer is calculated according to the earth's magnetic field, but because the geomagnetic field is easily disturbed, the calculated yaw angle is not too accurate. In order to solve the above problems, multi-sensor information fusion is widely used to estimate the attitude angle. In this paper, based on the previous fusion algorithms, our own fusion scheme is proposed: the Kalman filter algorithm is applied to the three-axis gyroscope. The three-axis accelerometer and three-axis magnetometer are fused to obtain the optimal angle estimation. Firstly, the random drift AR error model is established according to the output data of the sensor, and the error compensation and filtering of the original data are made. Then, the angle of the gyroscope calculation is used as the estimation angle. The angle calculated by accelerometer and magnetometer is taken as the measuring angle, and the optimal attitude angle estimation is obtained by combining the estimation angle with the measurement angle using Kalman filter algorithm. The proposed fusion algorithm is verified to achieve the desired results. Finally, according to the designed algorithm, the attitude angle detection device which can be worn on the waist is designed by using MPU-9150 and Arduino Pro mini, which can be worn in the waist. The reliability of the device is proved by the actual test and the improved advice is put forward. In this paper, as an engineering application, attitude angle estimation is applied to the detection of human fall attitude, which plays an important role in the application of attitude estimation in other fields, and has certain application value and practical significance.
【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN713;TP212
【参考文献】
相关期刊论文 前6条
1 金光明;张国良;陈林鹏;田琦;;MEMS陀螺仪静态漂移模型与滤波方法研究[J];传感器与微系统;2007年11期
2 王森;陈晓冬;杜承阳;汪毅;郁道银;;磁干扰环境下基于多传感器的内窥镜姿态定位技术[J];传感技术学报;2013年08期
3 吕印新;肖前贵;胡寿松;;基于四元数互补滤波的无人机姿态解算[J];燕山大学学报;2014年02期
4 臧荣春,崔平远;陀螺随机漂移时间序列建模方法研究[J];系统仿真学报;2005年08期
5 田蔚风,金志华,陆恺;神经网络在陀螺漂移误差模型辨识中的应用[J];中国惯性技术学报;1998年03期
6 刘鲁源,陈玉柱,陈刚,李士心;基于小波变换的陀螺漂移建模与实验研究[J];中国惯性技术学报;2004年01期
，本文编号：2214556
本文链接：https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2214556.html
