Detection Technology of Road Marks Utilizing Combination of Partial Templates

Abstract

To detect road marks as the references of localization for automated driving technologies, we developed a detection system which is robust against disturbances such as shadows, degradation, occlusions, and changes of road surface appearances due to different weather conditions. For the accurate distance estimation, our system used the “camera-down system”; furthermore, for making the most of it, our system used the combinations of multi partial templates which focused on different parts of road marks. From evaluations of actual driving data on a prefectural road in several weather conditions, we confirmed that our proposed system has the practical performances for correctly detecting road marks including disturbed ones.