使用MEMS惯性测量单元和拓扑地图的室内定位与导航技术研究

发布时间：2018-03-13 04:39

  本文选题：室内定位与导航　切入点：MEMS传感器　出处：《电子科技大学》2015年硕士论文　论文类型：学位论文

【摘要】：室内定位与导航潜藏着巨大的社会功能和经济效益,鉴于GNSS导航在室内环境中存在严重的信号衰减和多径问题,使用惯性测量单元进行不依赖于外部环境的自主定位与导航技术成为该领域研究的热点。本文即使用MEMS惯性测量单元装配在行人脚部进行行走动作的测量,并作航位推算,然后结合行人所在室内的拓扑地图估计其在室内的位置和航迹。具体而言,本文在研究和构建完整的基于MEMS惯性测量单元的室内定位与导航试验系统中完成如下工作:首先,本文针对室内定位与导航领域对精度、实时性和成本等各方面的需求,对导航系统平台的架构和整体工作流程做了设计,并针对其中所涉及的MEMS惯性测量传感器精度级别较低的特点进行了误差分析,在对惯性测量传感器的误差分析中,将其测量误差源分为确定性偏差与随机噪声两类,提出了一个综合性的校准方案。其次,传统的行人航位推算方法基于扩展卡尔曼滤波器和对行人脚步站姿的检测方法从而得到最优估计的行人航迹。但这一方法并未考虑惯性测量传感器的确定性偏差,使得推算航迹偏离实际航迹较大。本文所提出的最优航迹推算方案加入了对传感器确定性偏差的校准,通过建立一个综合性的误差模型,将计算得到的确定性校准系数以及Allan方差作为参数用于卡尔曼滤波器中,同时对两类误差进行补偿,根据所使用的传感器,以实际行走总距离为基准,新的方案使得最终航迹推算的位置精度相比传统方法提高了30%左右。然后,本文根据行人航位推算结果使用基于粒子滤波的地图匹配算法估计行人在室内的位置与航迹。这一地图匹配算法需要利用到室内地图中门、房间、走廊等结构信息及其空间关系,本文采用一种容差性的方案提取所需建筑结构并构建室内拓扑地图。不同于已有室内地图拓扑化方案中将走廊作为路径,走廊中拐点作为节点的拓扑数据结构,本文在提取出室内门、房间、走廊等空间结构后,将可以通行的空间作为路径以及将门作为节点进行室内拓扑地图的构建,一方面可以使用可通行空间的轮廓作为路径约束,用于地图匹配所使用粒子滤波算法的重要性采样中,另一方面将房间也纳入到可通行空间中,更加符合实际中行人常常在房间之间进行活动的特点。最后,采用手机客户端实现演示系统,对地图匹配算法进行了实验验证,结果显示最终地图匹配的位置精度可以达到±1.5m左右,行人可以根据匹配结果确定当前所在室内的位置,证明了该室内导航方案的有效性。
[Abstract]:Indoor positioning and navigation have great social functions and economic benefits. In view of the serious signal attenuation and multipath problems in indoor environment, GNSS navigation has many problems. The use of inertial measurement units for autonomous positioning and navigation independent of the external environment has become a hot topic in this field. In this paper, the MEMS inertial measurement unit is assembled in the foot of the pedestrian to measure the walking motion, and the navigation calculation is made. Then, the location and track of the pedestrian are estimated by using the topological map of the indoor. Specifically, this paper studies and builds a complete indoor positioning and navigation test system based on the MEMS inertial measurement unit. The main work is as follows: first of all, Aiming at the requirement of precision, real-time and cost in the field of indoor positioning and navigation, this paper designs the architecture and overall workflow of the navigation system platform. The error analysis of the MEMS inertial measurement sensor is carried out according to its low precision level. In the error analysis of the inertial measurement sensor, the measurement error source is divided into two categories: deterministic deviation and random noise. A comprehensive calibration scheme is proposed. Secondly, The traditional footpath estimation method is based on the extended Kalman filter and the detection method of the footstep posture to obtain the best estimated pedestrian track, but the deterministic deviation of the inertial measurement sensor is not taken into account in this method. The optimal track calculation scheme proposed in this paper includes the calibration of the deterministic deviation of the sensor, and establishes a comprehensive error model. The calculated deterministic calibration coefficient and Allan variance are used as parameters in Kalman filter, and two kinds of errors are compensated. According to the sensor used, the actual total walking distance is taken as the reference. The new scheme improves the accuracy of the final track calculation by about 30% compared with the traditional method. Then, In this paper, we use particle filter based map matching algorithm to estimate the indoor position and track of pedestrians according to the results of pedestrian reckoning. This map matching algorithm needs to be used in the door and room of the indoor map. In this paper, a tolerance scheme is used to extract the required building structure and construct the indoor topological map, which is different from the corridor as the path in the existing topologies of indoor maps. The inflection point in the corridor is the topological data structure of the node. After extracting the indoor door, room, corridor and other spatial structures, the passable space is taken as the path and the door is used as the node to construct the indoor topological map. On the one hand, the contour of the passable space can be used as the path constraint to sample the importance of the particle filter algorithm used in map matching, on the other hand, the room can also be incorporated into the passable space. Finally, the mobile phone client is used to implement the demonstration system, and the map matching algorithm is verified experimentally. The results show that the position accuracy of the final map matching can reach 卤1.5 m, and the pedestrian can determine the current indoor location according to the matching results, which proves the validity of the indoor navigation scheme.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP212;TN713
，

本文编号：1604877