基于隐式众包Wi-Fi的航站楼大型室内高精度定位关键技术研究

In response to the rapid growth of the future civil aviation passenger flow and the personalized demand for location based services, the smartphone-based Wi-Fi fingerprinting of airport terminal high-accuracy indoor localization technology as a new generation of information construction for civil aviation development direction has attached great attention，while the relatively fundamental research is very meaningful. To avoid the labor-intensive and time-consuming in site survey, we propose to utilize implicit crowdsourcing-based data collection. To mitigate the heterogeneous environment coursed by device and human activity, the use of kernel estimation and the relative strength of received signal strength (RSS) value is proposed. In accordance with the problem that the extraction of fingerprint data for the mapping process to the actual location, we aim to find the feature point in the fingerprints, and research the key technology of “feature-based skeleton graph matching” for step-by-step mapping. In terms of the unknown AP location and only RSS data available, we propose to break through the key technology of trajectory-constraint, location features and sensors according to the characteristics of user motions. Accordingly, the localization accuracy for the large area of airport terminal can be improved. By solving the corresponding key issues, we lay a solid theoretical and technical foundation for developing new high-accuracy indoor localization system in an airport terminal.

为应对未来民航客流量的迅猛增长和旅客对位置服务的个性化需求，基于智能手机Wi-Fi的航站楼室内高精度定位技术作为新一代民航信息化建设的发展方向，受到人们广泛关注，相关的基础性研究具有重要的科学和实际意义。本项目拟采用隐式众包模式降低指纹数据采集所需的大量开销。针对设备异构性问题，提出利用核函数估计和RSS相对强弱的方法；针对指纹数据的提取并与实际标定位置映射的问题，拟开展特征骨骼图匹配的研究，实现分步骤、分阶段的映射；对于AP位置未知且无其它外设标签辅助的条件下，拟综合利用连续轨迹、指纹特征点、传感器信息等多层约束来提高现有算法的定位精度。通过解决相应关键基础问题，为大型室内高精度定位系统的研制奠定坚实的理论和技术基础。

本项目面向未来民航客流量的迅猛增长和旅客对位置服务的个性化需求，研究了基于智能手机Wi-Fi的航站楼室内高精度定位技术，通过采用隐式众包模式降低指纹数据采集所需的大量开销。针对设备异构性问题，提出了利用核函数估计和RSS相对强弱进行异构性抑制；针对指纹数据的提取并与实际标定位置映射的问题，研究了特征骨骼图匹配的方案，实现分步骤、分阶段的映射；对于AP位置未知且无其它外设标签辅助的条件下，综合利用连续轨迹、指纹特征点、传感器信息等多层约束来提高现有算法的定位精度。进一步，尝试通过同时定位与制图(SLAM)技术进行地图的自动绘制与构建，避免繁琐的人工采集操作。最后，通过搭建机器人操作系统(ROS)以及手机APP上的实际测试，发现利用众包数据采集后定位精度相对于专业人员采集后的定位精度有少量损失。如果引入 ROS进行大量自动数据采集则可以有效弥补定位精度的损失。通过解决这些关键问题，为大型室内高精度定位系统的研制奠定坚实的理论和技术基础。