Institute for
Robotics and Process Control

Deutsch   English

Camera-based lane border detection



Autonomous navigation as well as assisting the driver in safely driving a vehicle requires the correct recognition and interpretation of the vehicle environment. This includes the detection of the drivable area in front of the vehicle or its boundary - the lane border.
Obstacles such as cars and curb stones not lying on the ground plane can be extracted from 3D data. These can be obtained e.g. by stereo cameras, laser or ultrasonic sensors. In many cases, however, a purely image-based lane border detection is required.

Project description

Within the Volkswagen research project "eT" (electric transporter), a post van was developed to facilitate the work of the postman. The vehicle can automatically follow the postman along the roadside, while he is walking on the sidewalk. Thus between close delivery locations the postman doesn't have to get back into the car to drive the short distance to the next destination.
For this purpose a camera-based lane border detection has been developed which allows the vehicle to automatically drive along the lane border in arbitrarily structured environments.


Results of the lane border detection (plotted in red)


Video clip from the eT presentation in November 2011 (YouTube)(External)

It took 0.28s to generate this page.