Real Time Lane Detection in Urban Streets

Description

This project was part of Team Caltech, Caltech's entry in the DARPA Urban Challenge in November 2007. This project's main aim was to detect and localize lane lines in urban streets, which will help Alice, Team Caltech's autonomous vehicle, find its way in traffic.
The system was implemented in C++ in over 10,000 lines of code. It achieved real time performance of about 50Hz and average accuracy of 95%.

Results

The videos below show detection results on the four video clips for the following two detection modes:

• Mode-all: detecting all lane markers on the street
  High quality [63 MB] or Low quality [19 MB]
• Mode-2: detecting only the markers of the current driving lane
  High quality [64 MB] or Low quality [18 MB]

Mode-all Movie

Mode-2 Movie

Software

• Caltech Lane Detection Software: C/C++ package that implements the lane detection algorithm in [1] and that produced these results.
• CaltechCV Image Labeler: a Matlab script to edit/view the labels in the Caletch Lanes Dataset.

Dataset

Caletch Lanes Dataset.

References

1. Mohamed Aly, Real time Detection of Lane Markers in Urban Streets, IEEE Intelligent Vehicles Symposium, Eindhoven, The Netherlands, June 2008. [pdf]
2. Mohamed Aly, Joel W. Burdick, Vanessa Carson, Stefano Di Cairano, Laura Lindzey, Jeremy Ma, Richard M. Murray, Richard Petras, Sam Pfister, Dominic Rizzo, Tichakorn Wongpiromsarn. Sensing, Navigation and Reasoning Technologies for the DARPA Urban Challenge, Technical Report, Caltech, USA, 2008. [pdf]