基於點雲之高精地圖車道線自動萃取研究

LANE LINE EXTRACTION FOR HD MAPS BASED ON POINT CLOUD

10.6652/JoCICHE.202110_33(6).0001

曾芷晴(Jhih-Cing Zeng)；蔡光哲(Guang-Je Tsai)；江凱偉(Kai-Wei Chiang)

自動駕駛 ； 高精地圖 ； 點雲 ； 車道標線 ； HD Maps ； autonomous vehicle ； mapping regulation ； feature extraction

中國土木水利工程學刊

33卷6期（2021 / 10 / 01）

409 - 415

繁體中文

近年來自動駕駛技術持續穩定發展，並深受國內外業界與學界重視，而高精地圖在自動駕駛技術發展上扮演不可或缺的角色，即便機器擁有遠優於人腦的計算能力與效率，但是仍缺乏行使決策的能力，且需經過一連串複雜的計算處理才能回應當下環境的刺激，因此必須為自駕車量身打造一套新的地圖，即為高精地圖。高精地圖不僅能提供豐富的三維環境空間資訊，也能在導航過程中提供先驗資訊，交通道路元素的屬性，如車道邊界、交通號誌、道路幾何關係，都被精確定義於高精地圖中，因此高精地圖為自動駕駛技術提供額外的輔助資訊。一般而言，自駕車或測繪車上會搭載許多感知感測器，如光達、相機，以快速有效地收集周遭精確的空間資訊，然而後續地圖建置的過程仍仰賴人工處理，此過程非常耗費時間與人力成本，因此本研究使用移動式測繪系統所攫取的資料中自動萃取高精地圖中的特性道路元素，由於點雲具備精確的三維位置的優勢，因此本研究著重於從點雲資料中獲取萃取控車所需的道路環境特徵，即車道標線，以建置高精地圖。為了降低資料處理所花費的運算時間並得到較好的成果，本研究利用高程濾波器過濾點雲，保留地面點雲資料，並將道路點雲資料體素化（voxelization），依據道路與車道標線幾何以及點雲的回波特性，萃取車道標線，最後對萃取成果進行演算法的可行性分析與評估。

Reacting to the era of autonomous vehicle, various domestic and foreign experts and companies have been actively involved in related studies and development. High Definition Maps (HD Maps) also play an indispensable role in the development of autonomous driving technology. Although robots have Even though the machine has much better computing ability and efficiency than the human brain, it still lacks the ability for decision-making and judgment. On the other hand, it still needs to conduct a series of complex computing to react to the surrounding changing and emergent event. Therefore, it is essential to construct the exclusive maps, HD Maps, for machine to realize the autonomous driving. HD Maps not only provide rich 3D geospatial information but also provide a prior information during navigating. The attribute of road elements, such as lane line, centerline, traffic line, road geometry, are precisely defined in HD Maps. Hence, HD Maps can provide additional environment information for autonomous vehicle. In general, there are several mapping sensors, such as LiDAR and camera, mounted on the autonomous vehicle and commercial Mobile Mapping System (MMS) to efficiently collect the accurate geospatial information. However, the subsequent procedure of map generation relies on manpower. This procedure is time-consuming and laborious. Therefore, this study proposes to extract the certain road element in HD Maps from data collected by MMS. Sine point cloud has highly accurate 3D positioning information, this study focuses on point cloud processing to extract the important road elements for vehicle control. On the other hand, this study proposes to extract the lane line for HD Maps generation. In order to reduce the computational time and improve the results, this study conducts an elevation filter to retain the ground point cloud. Next, the voxelization is implemented on the ground point cloud to extract the feature of lane line based on the geometry and intensity information. Finally, the results are assessed to evaluate the feasibility and performance of the proposed algorithm.

1. Aly, M.(2008).Real time detection of lane markers in urban streets.2008 IEEE Intelligent Vehicles Symposium
2. Bogoslavskyi, I.,Stachniss, C.(2017).Efficient online segmentation for sparse 3D laser scans.PFG–Journal of Photogrammetry, Remote Sensing and Geoinformation Science,85(1),41-52.
3. Clode, S.,Kootsookos, P. J.,Rottensteiner, F.(2004).the Automatic Extraction of Roads from LIDAR Data.International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences,35(Part B3)
4. De Silva, V.,Roche, J.,Kondoz, A.(2018).,未出版
5. Guan, H.,Li, J.,Yu, Y.,Chapman, M.,Wang, C.(2014).Automated road information extraction from mobile laser scanning data.IEEE Transactions on Intelligent Transportation Systems,16(1),194-205.
6. He, B.,Ai, R.,Yan, Y.,Lang, X.(2016).Lane marking detection based on convolution neural network from point clouds.2016 IEEE 19th International Conference on Intelligent Transportation Systems (ITSC)
7. Ibrahim, S.,Lichti, D.(2012).Curb-based street floor extraction from mobile terrestrial LiDAR point cloud.International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences,39,B5.
8. Jiang, H.(2017).University of Waterloo.
9. Kumar, P.,McElhinney, C. P.,Lewis, P.,McCarthy, T.(2014).Automated road markings extraction from mobile laser scanning data.International Journal of Applied Earth Observation and Geoinformation,32,125-137.
10. Litman, T.(2017).Autonomous Vehicle Implementation Predictions.Victoria, Canada:Victoria Transport Policy Institute.
11. Smadja, L.,Ninot, J.,Gavrilovic, T.(2010).Global environment interpretation from a new mobile mapping system.2010 IEEE Intelligent Vehicles Symposium
12. Song, W.,Yang, Y.,Fu, M.,Li, Y.,Wang, M.(2018).Lane detection and classification for forward collision warning system based on stereo vision.IEEE Sensors Journal,18(12),5151-5163.
13. Stephenson, S.,Meng, X.,Moore, T.,Baxendale, A.,Ford, T.(2011).Accuracy requirements and benchmarking position solutions for intelligent transportation location based services.Proceedings of the 8th International Symposium on Location-Based Services
14. Vardhan, H., HD Maps: New Age Maps Powering Autonomous Vehicles, Geospatial World (2017).
15. Yang, B.,Fang, L.,Li, Q.,Li, J.(2012).Automated extraction of road markings from mobile LiDAR point clouds.Photogrammetric Engineering & Remote Sensing,78(4),331-338.
16. Yang, B.,Wei, Z.,Li, Q.,Li, J.(2012).Automated extraction of street-scene objects from mobile LiDAR point clouds.International Journal of Remote Sensing,33(18),5839-5861.
17. Yu, Y.,Li, J.,Guan, H.,Wang, C.,Yu, J.(2015).Semi-automated extraction of street light poles from mobile LiDAR point-clouds.IEEE Transactions on Geoscience and Remote Sensing,53(3),1374-1386.
18. Zhao, H.(2017).University of Waterloo.