Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : June 05, 2019 - June 08, 2019
In this paper, we describe accuracy evaluation of 3D point clouds applied to autonomous driving. An error due to measurement may remain in the constructed 3D point cloud. Also, in the case of using for localization, the problem that localization failed is known due to the place where the continuous structure. The purpose of this study is to examine the evaluation index of the map and evaluate the accuracy of the 3D point cloud. In the proposed method, we evaluate the map from the viewpoint of Localization and propose an evaluation method using convergence by matching. By the proposed evaluation method, it was found that the place where Localization is possible and the place where the error remained in map are clearly separable. Moreover, we think that the continuous structure place can be classified from the tendency of the convergence point.
