Section: New Results

LIDAR-based lane marking detection for vehicle localization

Participants : Farouk Ghallabi, Fawzi Nashashibi.

Accurate self-vehicle localization is an important task for autonomous driving and ADAS. Current GNSS-based solutions do not provide better than 2-3 m in open-sky environments. In order to achieve lane-level accuracy, a lane marking detection system using a multilayer LIDAR (velodyne) and a map matching algorithm has been introduced. The perception system includes three different steps: road segmentation, image construction and line detection. Our road segmentation method purely relies on geometric analysis of each layer returns. Detected lane markings are matched to a prototype third party map which was built with absolute accuracy = 5cm. The map matching algorithm is a particle filtering process that achieves lane-level accuracy (20 cm). More details are in [23]. This work has been partially funded by Renault.