Conference Proceeding
Stereo road detection based on ground plane
Autor
Rodríguez-Garavito, C. H.
Carmona-Fernández, J.
de la Escalera, A.
Armingol, J. M.
Institución
Resumen
This paper presents a robust road perception algorithm aimed to detect multiple lanes with temporal integration, one of the most important tasks in Advanced Driver Assistance Systems (ADAS). A new vision-based system is proposed, consisting on three parts: a line marker detection algorithm, a road line classification and a lane tracking integration. The goal is to detect the position, type and number of road lanes. The developed approach is characterized by the use of the bird’s eye view, road marks filtering based on gradient space algorithms, robust features descriptor for line classification, and road tracking based on time of life for each detected lane. The road detection is done according to the Spanish standard IC 8.2. The system was tested on the test platform IvvI 2.0.