全息高精度导航地图:概念及理论模型

本文提出了全息高精度导航地图的概念,它融合了多源数据,尤其是电磁传感器、声音传感器、热红外仪等传感器数据,从更多角度为导航提供信息。在此概念基础上,提出了一种融合多源数据的全息高精度导航地图理论模型框架,该框架包含4个步骤:①全息道路数据采集,包含道路三维彩色激光点云、遥感影像、无人机航拍倾斜测量数据、摄像头图像、热红外图像、声场信号、电磁场信号;②道路静态信息提取,从上述采集信息提取和标记得到,如车道线、路坎、栏杆、路牌、路灯、隧道等,作为无人驾驶车辆规划基础路线和车辆位置定位的数据基础;③道路动态信息提取,从上述采集信息提取和标记得到,如离前后车辆的距离、前方有无行人、道路施工护栏、泛在信息等,作为检测无人驾驶车辆周围的实时道路环境和规划无人驾驶车辆行驶路线的依据;④动、静态信息融合:融合道路静态信息和道路动态信息,丰富道路信息,提高道路线精度,提高全息高精度导航地图更新的效率,为导航和无人驾驶车辆提供地图服务。

Pan-information-based High Precision Navigation Map: Concept and Theoretical Model

Map is intelligent product of human civilization. The rapid development of science and technology increased the diversity and readability of map. Based on the existing concepts and forms of map, this paper developed a concept of pan-information-based high precision navigation map.This new map concept was a kind of intelligent navigation map which was oriented to vehicle mobile operation and integrated more functions, such as environment perception, pan-features fusion, high-precision positioning and planning decision-making. It can collect and fuse different information based on unified data model for different application fields, and it was a brand-new map form. Key innovation of this map was capturing and fusing pan-information of road from multi-source sensors, especially electromagnetic sensors, sound sensors, thermal infrared instruments and the others, in order to provide information for navigation from more aspects. Based on this concept, a theoretical model framework of pan-information-based high precision navigation map with multi-source data fusion was proposed. This framework consisted of four parts: (1) Pan-information road data collection. Data acquisition vehicles and other external sensors were used to collect multi-source data such as LiDAR system point cloud data, remote sensing images, oblique photogrammetric data, high-definition camera images, thermal infrared images, sound signal and electromagnetic signal. (2) Road static information extraction. Static information was basis for route planning and vehicle locating, which was obtained through the multi-source data mentioned above. The main road static information included lane lines, curbs, railings, road signs, road lamps, tunnels. (3) Road dynamic information extraction. Dynamic information was basis for real-time detecting surroundings and adjusting route of auto-vehicle, which was also obtained by extracting and marking the above collected data. The main road dynamic information included the distances between the vehicle and near objects, such as other cars, pedestrians and construction guardrails. Road dynamic information also included some ubiquitous information such as meteorological data, dynamic traffic conditions, POI data. (4) Fusion of dynamic and static information. Integrating road static information and road dynamic information can enrich road information, increase the accuracy of lane line, improve the updating efficiency of pan-information-based high precision navigation map,and provide map services for auto-driving vehiclesand its navigation. Compared with the existing map concepts and technologies, the map proposed in this paper has two characteristics: more detailed road information and more efficient data update, and both of the characteristics were based on our richer data sources, more diverse data collection methods and more efficient information extraction algorithms.