基于众包IMU数据的室内地图建立方法研究

发布时间：2019-05-14 07:05

【摘要】：如今,随着智能手机的迅速普及,高精度的室内定位服务正在快速的发展,人们可以使用随身携带的智能手机来享受定位服务。一般来说,无论是基于WiFi,蓝牙,zigbee还是视觉等定位技术,它们的部署都需要提前获取精确的室内地理信息。在精确的室内地理信息帮助下,室内导航系统可以在离线阶段构建特定的数据库,并且当在线阶段返回定位结果时,能够在智能手机中清楚地显示其位置和导航估计。但是,如果在未知环境或未获得准确的室内建筑图的情况下,缺少实现室内定位技术部署的先决条件,因此需要一种低成本的快速建立室内地图技术。因此针对在未知室内环境缺少室内地图的问题,本文提出了一种众包室内地图建立方法,实现了在未知环境下通过众包用户IMU(Inertial measurement unit)数据,完成室内地图的快速建立。具体研究内容如下:首先针对PDR(Pedestrian Dead Reckoning)算法在众包测量中精确性较差的问题,提出了一种基于PDR算法的众包用户姿势识别与轨迹起始点检测算法,该算法能够有效的在不同测量模式下生成众包PDR轨迹,并进一步提升轨迹精度。其次,针对室内用户轨迹分布缺少统一模型的问题,根据室内用户行走习惯,建立了一种室内环境用户行走模型,该模型能够有效的描述用户在室内行走习惯并给出PDR轨迹的概率分布。另外,本文通过一种基于密度峰值的聚类算法对长直走廊数据进行聚类,并根据3?原理,计算得到每段长直走廊对应的走廊宽度,获取了室内地图建立的重要参数。最后,针对室内众包用户轨迹构建室内地图时存在PDR误差与众包误差的问题,本文基于室内行人行走模型,提出了一种基于轨迹密度分析的室内地图建立方法,该方法能够有效的在未知环境下构建室内地图。该方法包括对热点区域划分,轨迹密度计算,热点区域筛选,地图轮廓生成与直线化地图生成。本文利用Alpha-shape算法对保留的热点区域进行边缘提取,得到室内地图轮廓。最终通过求解得到的走廊宽度进行轮廓的直线化生成,得到精确的室内地图。通过与真实环境室内地图对比,本文提出的基于众包IMU数据的室内地图建立方法能在保证精度的条件下,实现室内地图的低成本快速建立。从而能够在未知环境下实现快速获取室内地理信息,扩展了室内定位的应用前景。
[Abstract]:Nowadays, with the rapid popularity of smartphones, high-precision indoor positioning services are developing rapidly, people can use portable smartphones to enjoy positioning services. Generally speaking, whether based on WiFi, Bluetooth, zigbee or visual positioning technology, their deployment needs to obtain accurate indoor geographic information in advance. With the help of accurate indoor geographic information, the indoor navigation system can build a specific database in the offline phase, and when the positioning results are returned in the online phase, the location and navigation estimates can be clearly displayed in the smartphone. However, if the environment is unknown or the accurate indoor building map is not obtained, there is no prerequisite for the deployment of indoor positioning technology, so it is necessary to establish indoor map technology quickly and cheaply. Therefore, in order to solve the problem of lack of indoor map in unknown indoor environment, this paper proposes a method of building indoor map in crowdsourcing, which realizes the rapid establishment of indoor map through crowdsourcing user IMU (Inertial measurement unit) data in unknown environment. The specific research contents are as follows: firstly, in order to solve the problem of poor accuracy of PDR (Pedestrian Dead Reckoning) algorithm in crowdsourcing measurement, a crowdsourcing user posture recognition and trajectory starting point detection algorithm based on PDR algorithm is proposed. The algorithm can effectively generate crowdsourcing PDR trajectories in different measurement modes, and further improve the trajectory accuracy. Secondly, according to the walking habits of indoor users, a kind of indoor environment user walking model is established to solve the problem of lack of unified model for indoor user trajectory distribution. The model can effectively describe the indoor walking habits of users and give the probability distribution of PDR trajectory. In addition, this paper uses a clustering algorithm based on density peak value to cluster the long straight corridor data, and according to 3? In principle, the corridor width corresponding to each long straight corridor is calculated, and the important parameters of indoor map establishment are obtained. Finally, in order to solve the problem of PDR error and crowdsourcing error when indoor crowdsourcing user trajectory is constructed, based on indoor pedestrian walking model, an indoor map establishment method based on trajectory density analysis is proposed in this paper. This method can effectively construct indoor map in unknown environment. The method includes the division of hot spots, the calculation of trajectory density, the screening of hot spots, the generation of map contours and the generation of straight maps. In this paper, Alpha-shape algorithm is used to extract the edge of the reserved hot spot, and the outline of indoor map is obtained. Finally, by solving the width of the corridor, the outline is straightened and the accurate indoor map is obtained. Compared with the real environment indoor map, the indoor map establishment method based on crowdsourcing IMU data can realize the low cost and fast establishment of indoor map under the condition of ensuring the accuracy. Thus, the indoor geographic information can be obtained quickly in unknown environment, and the application prospect of indoor positioning can be expanded.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN96

【参考文献】