Automated Modeling of Road Networks for High-Definition Maps in OpenDRIVE Format Using Mobile Mapping Measurements

Geomatics

Naser El-Sheimy¹  Kai-Wei Chiang²  Jhih-Cing Zeng²  Meng-Lun Tsai²  Hao-Yu Pai² 

[1] Department of Geomatics Engineering, The University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada;Department of Geomatics, National Cheng Kung University, Tainan City 701401, Taiwan;

关键词: HD maps;    OpenDRIVE;    extraction of road factors;    lane lines;    point clouds;    automated modeling;   

DOI  :  10.3390/geomatics2020013

来源: DOAJ

【 摘 要 】

With growing attention being devoted to autonomous vehicle (AV) safety, people have recently attached importance to high-definition (HD) maps. HD maps are not limited by environmental factors and can limit AVs driving in certain lanes. HD maps provide accurate auxiliary information on factors such as road geometry, traffic sign placement, and traffic topology. Nowadays, most HD maps are made from point clouds data, and this data contains accurate 3D position information. However, the production costs associated with HD maps are substantial. This article proposes an algorithm that reduces a great amount of time and human resource. The algorithm is divided into three phases, lane lines’ extraction from point clouds, modelling lane lines with attributes, and building OpenDRIVE file. The algorithm extracts lane lines resting on intensity value within the range of roads. Next, it models lane lines by cubic spline interpolation with the result of first phase, and build the OpenDRIVE file following the announcement of OpenDRIVE. The final result is compared with the verified HD map from the mapping company to analyze the accuracy. The root mean square (RMSE) obtained were 0.069 and 0.079 m for 2D and 3D maps, respectively.

【 授权许可】

Unknown