基于MEMS传感器的高精度行人导航算法研究

发布时间：2019-03-31 00:11

【摘要】：伴随微电子技术迅猛发展,导航技术的应用领域已渐渐从军事化转向了商用化。近几年,室内导航技术得到了大力发展,受到了越来越多企业和民众的重视。但现有的室内定位系统多是基于基础设施实现,其缺点非常明显：需要提前在需要导航定位的环境中安装大量外部设备,投入成本不菲,且精度仍有待提高。本文拟设计一种基于自身传感器的行人导航系统来解决上述缺陷。捷联式惯性导航技术不仅可以避免需要外设、投入高等缺陷,并且在短时间定位精度相当高,但随着时间的推移,其存在累积的漂移误差。针对这一点,本文基于卡尔曼滤波设计出一种有效补偿累积误差的算法。首先,通过捷联式惯性计算模块对惯性测量信息进行积分解算,由角速度积分求解出姿态信息,通过四元数法对加速度进行坐标转换,而后积分求解出速度信息,再二次积分求解出位置信息；然后,由零速检测模块通过三条件判断法对行人步行时的“零速度”阶段进行检测,当检测到“零速度”时触发卡尔曼滤波模块；最后,将捷联式惯性计算后的速度向量作为量测值,使用卡尔曼滤波来估计系统状态误差,通过求解出的速度估计协方差,对状态误差估计进行分段,融入后向固定区间平滑技术,进一步调整状态误差估计及其协方差矩阵,前向反馈校正行人位置、速度和姿态信息；同时,扩充了卡尔曼滤波状态模型,加入加速度计和陀螺仪的零偏误差信息,通过反馈校正惯性测量信息,进一步消除系统漂移误差,最终实现了室内环境的行人定位导航。在上述误差补偿算法的基础上,本文设计了一种行人导航系统,通过多组实验仿真验证,在500m路程内,该行人导航系统收敛在稳定状态,定位精度保持在1m之内,实现了高精度的定位目标。本文设计的行人导航系统,为基于自身传感器的室内定位提供了一种简单而有效的方法,可以进一步融合完善,并推广应用到智能家居、火场救援、超市购物、车库停车等诸多场景之中。
[Abstract]:With the rapid development of microelectronics technology, the application field of navigation technology has gradually changed from militarization to commercialization. In recent years, indoor navigation technology has been greatly developed by more and more enterprises and people. However, most of the existing indoor positioning systems are based on infrastructure, and their shortcomings are very obvious: a large number of external devices need to be installed in advance in the environment where navigation and positioning is needed, the cost of investment is very high, and the precision still needs to be improved. This paper proposes to design a pedestrian navigation system based on its own sensor to solve the above shortcomings. Strapdown inertial navigation technology can not only avoid the need for peripherals, put in high defects, and in a short time positioning accuracy is quite high, but with the passage of time, it has accumulated drift error. In view of this, this paper designs an effective algorithm to compensate cumulative error based on Kalman filter. Firstly, the inertial measurement information is solved by the strapdown inertial calculation module, the attitude information is obtained by the angular velocity integral, the acceleration is transformed by the quaternion method, and the velocity information is obtained by the integration. Secondly, the position information can be solved by quadratic integration. Then, the zero-speed detection module is used to detect the "zero-speed" phase of pedestrian walking through three-condition judgment method, and the Kalman filter module is triggered when the "zero-speed" is detected. Finally, the velocity vector calculated by strapdown inertia is used as the measurement value, and Kalman filter is used to estimate the state error of the system. By solving the covariance of velocity estimation, the state error estimation is segmented. Further adjust the state error estimation and its covariance matrix with backward fixed interval smoothing technology, and forward feedback corrects pedestrian position, velocity and attitude information. At the same time, the Kalman filter state model is extended, the zero bias error information of accelerometer and gyroscope is added, and the inertial measurement information is corrected by feedback to further eliminate the drift error of the system. Finally, the pedestrian positioning and navigation in indoor environment is realized. Based on the error compensation algorithm mentioned above, a pedestrian navigation system is designed in this paper. It is verified by many experiments that the pedestrian navigation system converges to a stable state and the positioning precision is kept within 1m within 500m distance. The high precision positioning target is achieved. The pedestrian navigation system designed in this paper provides a simple and effective method for indoor positioning based on its own sensor, which can be further integrated and applied to smart home, fire rescue, supermarket shopping, Garage parking and many other scenes.
【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

【相似文献】