Evaluation of High Precision Map Creation System with Evaluation Items Unique to Each Feature Type

Abstract

Nowadays, autonomous driving systems are being developed and utilized. And high precision 3D map data is said to be necessary to realize autonomous driving. This data contains position data of features such as “road line”, “road edge” and “road sign”. This map data is created with point cloud data taken by Mobile Mapping System (MMS), and to reduce cost, it is desirable that the map creation work is automated, but in fact it is difficult to fully automate. So first, features detection process will be performed by an automatic process, and then a manual confirmation process will be performed. Therefore, in order to reduce cost, it is important to improve a manual confirmation process. In this paper, we propose a system that improves a manual confirmation process by displaying features in different color using evaluation items unique to each feature type. In experiment, we used actual MMS data on Tokyo Japan, performed auto features detection process and got features data with evaluation item values. We investigated whether an appropriate low evaluation item values is given for error detection. As a result, we confirmed the effectiveness of this system.