基于情景感知的可靠可信城市车辆定位关键技术研究

With the booming of modern transportation in China, automobile has greatly affected every aspect of the public's daily life and business. The location information of vehicles obtained in real time, especially in urban scenarios, is of great importance to driving safety, road efficiency and experience of travelers. Existing localization schemes such as GPS, radar and image processing are neither reliable nor accurate for localizing vehicles in urban environments. This project will focus on context-aware reliable and accurate vehicular localization in urban settings, which will be directed in the following three thrusts. First, we will investigate new context types suitable for vehicular localization and study the theory and feasible methods to determine the usability of those new context types. Second, we will further explore the temporal and spacial characteristics of those newly found context information and design vehicular relative localization schemes based on the correlation of context information. Furthermore, we will take the system scalability into account and study cost-efficient tracking approaches. Last, we will utilize the popularity of smart mobile devices such as smartphones and tablets to study methods on establishing large-scale urban context maps. Moreover, we will design new map matching algorithms based on heterogeneous high-demension context information of different types. The significance of the proposed research lies in the following respects: first, this project will help to push the frontier of traffic safety technology, ensuring travel safety; second, the cutting-edge research will help to establish more intelligent transportation systems; third, the project emphasizes validation of the protocols and algorithms via testbed experiment, which will enable the development of new products and help to foster a vital competitive edge for China industry in the international market place.

实时获取车辆精确位置信息，尤其是在城市环境下，对行车安全、交通管理和改善出行体验等应用十分重要。现有定位方法如卫星和基站定位、雷达和图像处理对于定位城市车辆并不可靠。本课题着重基于城市情景感知的可靠可信车辆定位关键技术研究。首先，拟通过被动监测和主动探查手段，探索新型车辆定位适用的情景上下文，提出可用性判定理论和方法，为研究基于情景上下文的定位技术奠定可靠的数据基础；其次，拟利用情景上下文轨迹信息的时空相关特性，研究基于情景上下文相关的车辆相对位置检测方法，考虑系统可扩展性，研究车辆追踪技术；最后，面向智能移动终端，研究城市大规模广义情景上下文地图构建方法，面向多源、异构、高维的情景上下文信息，研究广义情景地图匹配方法。本课题有科学的研究方案作支撑，坚实的研究基础和研究条件作保障。课题的成功开展将突破全天候、全场景的车辆定位的若干关键技术，为推进我国智能交通领域发展做出创新性的贡献。

位置信息是智能交通和智慧城市等系统应用的重要基础信息，针对城市环境卫星定位性能不稳定问题，本课题开展了基于泛在移动网络信号的可靠可信车辆定位与相互定位技术研究，具体在三个方面展开研究并取得预期成果：1）课题组搭建了车载移动网络信号测量平台，对多个城市的移动蜂窝网络2G/3G信号进行了长时间、大规模的测量，对GSM-900的194个信道信号强度数据进行分析，发现结合了高维移动网络信号强度测量的车辆移动轨迹具有良好的时间稳定性、空间唯一性和高分辨率三个特性，可以作为定位使用的指纹特征；2）提出了基于复合轨迹相关性检测的车辆定位UPS和相互定位RUPS方法，在复杂城市环境下90%的定位精度为5.3米，比GPS系统精度高5倍；3）提出了大规模城市定位指纹数据库的构建方法，利用群智感知数据，实现低成本、高精度、动态更新的城市级定位指纹数据库。课题的研究形成了全天候、全场景的车辆定位关键技术，为推进我国智能交通领域发展做出创新性的贡献。