| 摘要: |
| 地磁基准图的构建是实现地磁匹配导航的基础。在室内、道路等地磁导航应用场景中，基于普通手机的众包建图技术是未来解决大规模地磁建图问题的可行手段，而众包数据位置坐标的获取是其中的关键环节。目前，众包数据位置的获取主要依靠手机自身全球卫星导航系统（GNSS）的定位结果，在室内等无GNSS信号的场景中主要通过惯性航位推算获得，但是存在城市峡谷区域GNSS定位结果偏移、室内长时间惯性航位推算易产生累积误差等问题，使得众包数据位置坐标精度难以支撑准确的地磁基准图构建需要。针对上述技术挑战，提出了一种基于地磁序列与开放街道地图（OSM）匹配的手机众包轨迹图优化方法。首先，通过GNSS/MEMS/OBD多源融合产生的初始轨迹构建相邻位姿点约束。其次，利用磁场的空间稳定特性，通过基于动态时间规整（DTW）的地磁序列匹配算法构建磁场闭环点约束。然后，利用OSM的绝对位置信息，通过基于隐马尔可夫模型（HMM）的地图匹配算法构建地图匹配点约束。最后，基于上述3类约束构建联合图优化模型，并在通用图优化（G2O）框架下通过列文伯格-马夸尔特（LM）算法获得众包轨迹的室内外全局优化结果。在城市峡谷和地下车库等弱或无GNSS场景下进行室内外连续众包数据采集和轨迹优化测试，平均定位均方根误差降低了54%，显著优于原始GNSS或GNSS/MEMS/OBD多源融合的定位结果。 |
| 关键词: 众包轨迹优化 图优化 众包建图 地磁序列 地图匹配 |
| DOI： |
|
| 基金项目:中国科学院青年创新促进会基金(E13314010D) |
|
| Optimization of smartphone crowdsourced trajectory maps based on geomagnetic sequence and OSM map matching |
| HONG Zilin,LI Wen,XIA Yupeng,WEI Dongyan,SHEN Ge,SONG Xinhang |
| (Aerospace Information Research Institute, Chinese Academy of Science, Beijing 100094, China; School of Electronic, Electrical and Communication Engineering, University of Chinese Academy of Science, Beijing 100049, China;Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China) |
| Abstract: |
| The construction of a geomagnetic reference map is essential for geomagnetic matching navigation. In indoor and road geomagnetic navigation scenarios, the crowdsourced mapping technology based on ordinary smartphones is a feasible solution for large-scale geomagnetic mapping in the future, and obtaining the position coordinates of the crowdsourced data is one of the key links. At present, the acquisition of the crowdsourced position data depends primarily on the positioning results provided by the smartphone's built-in global navigation satellite system (GNSS) function, and is supplemented by inertial dead reckoning in GNSS-denied environments, such as indoors. However, there are challenges, including GNSS signal drift in urban canyons and the accumulation of errors over time in indoor inertial navigation, which affect the accuracy of position data for precise geomagnetic mapping. To address these challenges, a smartphone crowdsourced trajectory optimization method based on geomagnetic sequence and OpenStreetMap (OSM) map matching is proposed. Firstly, adjacent pose point constraints are established by using initial trajectories generated by GNSS/MEMS/OBD multi-source fusion. Secondly, by exploiting the spatial stability characteristics of the geomagnetic fields, magnetic closed-loop point constraints are constructed by the dynamic time warping (DTW)-based geomagnetic sequence matching algorithm. Then, the absolute position information from OSM is used to create map matching constraints through a hidden Markov model (HMM)-based map matching algorithm. Finally, a joint graph optimization model incorporating these three constraints is constructed, and the indoor-outdoor global optimization results of the crowdsourced trajectories are achieved through the Levenberg-Marquardt (LM) algorithm under the general graph optimization (G2O) framework. The indoor-outdoor continuous crowdsourced data collection and trajectory optimization tests are carried out in GNSS-challenged environments, including urban canyons and underground parking lots, and demonstrate a 54% improvement in the average positioning root mean square error compared to raw GNSS or GNSS/MEMS/OBD multi-source fusion results, validating the effectiveness of the proposed method. |
| Key words: Crowdsourced trajectory optimization Graph optimization Crowdsourced map building Geomagnetic sequences Map matching |
