![]() However, the cumulative positional errors can quickly spiral out of control due to the long-term absence of global localization information, especially when compromised by inertial sensors with sub-par performance. With the information from inertial sensors such as gyroscope, accelerometers, or LiDAR odometry, the current position of the vehicle can be effectively inferred. ![]() ![]() Such GNSS signal degradation issues are dealt with by dead reckoning (DR). On typical urban roads, the accuracy of GNSS RTK struggles to remain stable due to satellite signals suffering from multipath fading and shadow effects by the nearby buildings and vegetation. Nevertheless, its high accuracy can only be maintained in open areas. The RTK technology can even achieve centimeter-level localization by modifying the mobile station with reference station data. GNSS is one of the most commonplace localization technology deployed on vehicles. For example, the vehicle must be able to localize itself before the information in the map becomes relevant for IV route planning in some applications such as the Road Experience Management of Mobileye, vehicle localization is a prerequisite for pending map updates. High-precision absolute vehicle localization is particularly important for HD map-based applications in IVs. In recent years, with the release of various high-precision map standards such as OpenDrive and NDS, vector HD maps from mainstream cartographers (e.g., TomTom and HERE) have thrived in commercial vehicles. It contains the precise geometric descriptions of road elements and retains higher-level semantic information such as road topology, lane type, speed limit, etc. On the other hand, Figure 1b shows an example of a vector HD map, of which the refined data are extracted from the raw point cloud map for a smaller data size. This kind of map retains raw geometric information and may be segmented with semantic labels however, it is not a market-favorable choice due to its huge data size. Some map-based applications implement the point cloud maps as shown in Figure 1a. Although HD maps are often deployed in costly equipment such as LiDAR and the corresponding high-performance integrated inertial navigation system, the HD map itself is not considered as a costly technology, and thus are widely available as an optional feature for most of the vehicles in the market. As a result, high-precision HD maps are considered the cornerstone of IV technology, especially for more advanced automated vehicles. ![]() With pre-collected environmental information, the HD map can act as a virtual sensor to improve vehicle safety without incurring additional hardware system complexity. In addition, we use low-cost on-board sensors and light-weight HD maps to achieve or even exceed the accuracy of existing map-matching algorithms.Īs an effective augmentation tool among on-board sensors, the HD compact map has gained tremendous popularity as a consumer vehicle add-on feature. Experiments have shown that MLVHM can achieve high-precision vehicle localization with an RMSE of 24 cm with no cumulative error. The effective data association method in MLVHM serves as the basis for the camera position estimation by minimizing feature re-projection errors, and the frame-to-frame motion fusion is further introduced for reliable localization results. The semantic features are delicately chosen for the ease of map vector alignment as well as for the resiliency against occlusion and fluctuation in illumination. Rooted upon these existing technologies, this article presents the concept of Monocular Localization with Vector HD Map (MLVHM), a novel camera-based map-matching method that efficiently aligns semantic-level geometric features in-camera acquired frames against the vector HD map in order to achieve high-precision vehicle absolute localization with minimal cost. The same cost-saving strategy also gives rise to an increasing number of vehicles equipped with High Definition (HD) maps. In the process of commercialization of IVs, many car manufacturers attempt to avoid high-cost sensor systems (e.g., RTK GNSS and LiDAR) in favor of low-cost optical sensors such as cameras. Real-time vehicle localization (i.e., position and orientation estimation in the world coordinate system) with high accuracy is the fundamental function of an intelligent vehicle (IV) system.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |