李必军, 郭圆, 周剑, 唐有辰, 董全华, 李治江. 智能驾驶高精地图发展与展望[J]. 武汉大学学报 ( 信息科学版), 2024, 49(4): 491-505. DOI: 10.13203/j.whugis20230287
引用本文: 李必军, 郭圆, 周剑, 唐有辰, 董全华, 李治江. 智能驾驶高精地图发展与展望[J]. 武汉大学学报 ( 信息科学版), 2024, 49(4): 491-505. DOI: 10.13203/j.whugis20230287
LI Bijun, GUO Yuan, ZHOU Jian, TANG Youchen, DONG Quanhua, LI Zhijiang. Development and Prospects of High Definition Map for Intelligent Vehicle[J]. Geomatics and Information Science of Wuhan University, 2024, 49(4): 491-505. DOI: 10.13203/j.whugis20230287
Citation: LI Bijun, GUO Yuan, ZHOU Jian, TANG Youchen, DONG Quanhua, LI Zhijiang. Development and Prospects of High Definition Map for Intelligent Vehicle[J]. Geomatics and Information Science of Wuhan University, 2024, 49(4): 491-505. DOI: 10.13203/j.whugis20230287

智能驾驶高精地图发展与展望

Development and Prospects of High Definition Map for Intelligent Vehicle

  • 摘要: 高精地图的发展在智能交通的推进中发挥着关键作用,是打造智能汽车与智慧城市“数据大脑”的基石,同时可推动智慧物流网络与交通安全风险监测网络的发展。从地图信息传输模型的角度出发,对已有的高精地图模型进行了分析,提出基于地图认知机理构建高精地图的思路,并对国内外高精地图格式标准进行了对比,提出了中国智能驾驶高精地图在标准编制时应遵循的原则。同时,从地图采集、处理与审核等维度展开,分析了当前高精地图生产与更新的关键技术,并通过车辆的规划与感知等应用实例,对高精地图的应用现状进行了分析。中国高精地图发展面临着模型复杂、生产更新能力难以满足发展需要、地理信息安全风险以及应用深度不足等挑战。针对以上挑战,所提出的基于认知模型建立高精地图闭环架构的思路,强调了保障数据安全、推动智能审图的重要性,指出了高精地图的广泛应用前景。

     

    Abstract:
    Objectives The development of high definition (HD) map is of paramount importance in advancing the digital infrastructure of transportation and serves as a fundamental component in creating the “data brain” of intelligent vehicles and smart cities. Furthermore, it promotes the growth of intelligent logistics networks and traffic safety risk monitoring systems.
    Methods The existing HD map models are analyzed from the perspective of map information transmission models, and the concept of constructing HD map models based on map cognition mechanisms is introduced. It also provides an overview of domestic and international standards for HD map formats. Moreover, the principles to be followed in the compilation of HD map for intelligent driving in China are outlined. Additionally, key technologies involved in HD map production and updates are explored, with a particular focus on map collection, processing, and verification. The current application status of HD map is analyzed through the examination of examples such as vehicle planning and perception.
    Results The development of HD map has made significant progress, but it also encounters difficulties and challenges. Existing HD map models tend to be complex and lack sufficient representation of dynamic elements, resulting in difficulties in maintaining real-time accuracy. Additionally, the production capacity of HD map remains insufficient to meet the demands of development, and challenges related to geographic information security and limited application depth persist.
    Conclusions A concept of establishing a closed-loop architecture for HD map using cognitive models is introduced. It highlights the significance of ensuring data security and promoting intelligent verification of maps and emphasizes the extensive application prospects of high-precision maps. During the research process of HD map, it is crucial to collaboratively address the difficulties and challenges, which can promote HD map to better serve intelligent transportation and societal development.

     

/

返回文章
返回