Skip to Main content Skip to Navigation

Détection d'obstacles multi-capteurs supervisée par stéréovision

Abstract : Road obstacle detection is a major topic for the developpment of future Advanced Driver Assistance Systems (ADAS). In this document, we propose a multi-sensor approach to obstacle detection. It is assumed to be robust and generic, thanks to a control task confered to stereovision. In the proposed approach, various sensors (stereoscopic sensor, laser scanner, optical identification sensor) provide hypothesis of detections, which are represented as volumes of interest in the disparity space associated to the stereoscopic images. These volumes are then processed by a stereovision algorithm to perform validation and characterization of the hypothesis. We propose a description of this methodology, three methods for the creation of hypothesis, and criteria for their validation. We also explore some practical aspects related to the implementation and the evaluation of the proposed strategy. Particularily, we propose to determine the appropriate method to obtain usable data from the stereoscopic images. Finally, three applications of the proposed methodology, fully operational in experimental vehicles, are presented. They figure how obstacle detection and cooperation with inter-vehicle communication could be used for the developpment of future ADAS.
Complete list of metadata

Cited literature [75 references]  Display  Hide  Download
Contributor : Mathias Perrollaz Connect in order to contact the contributor
Submitted on : Thursday, January 5, 2012 - 12:50:30 PM
Last modification on : Sunday, June 26, 2022 - 4:56:52 AM
Long-term archiving on: : Friday, April 6, 2012 - 2:31:00 AM


  • HAL Id : tel-00656864, version 1


Mathias Perrollaz. Détection d'obstacles multi-capteurs supervisée par stéréovision. Vision par ordinateur et reconnaissance de formes [cs.CV]. Université Pierre et Marie Curie - Paris VI, 2008. Français. ⟨tel-00656864⟩



Record views


Files downloads