The goal of this thesis has been to create a map of road lanes from piece-wisedetections. Using images from a stereo camera, an existing algorithm is able todetect lane tracks in real-time. With a basis spline (B-spline) representation ofthe tracks, these curve features are converted into point features. These are combinedoff-line with an odometric estimate for the creation of a motion and lanetrack map. Fusion of odometric and feature measurements is done using an optimizationbased simultaneous localization and mapping (slam) method. By use of visual inspection and a differential global positioning system (dgps) theresult is validated. The method is able to improve on existing motion estimatorsby using the curvature of lane tracks. A projection of the resulting map is usedto automatically label lane tracks as seen from camera view with accuracy andrange exceeding that of the real-time algorithm.
Automatic Lane Labeling Using an Off-line Smoothing Stereo-lane Algorithm
2017-01-01
Hochschulschrift
Elektronische Ressource
Englisch
DDC: | 629 |
Europäisches Patentamt | 2020
Stereo Augmented Detection of Lane Marking Boundaries
IEEE | 2015
|Prediction of Lane-Changing Maneuvers with Automatic Labeling and Deep Learning
Transportation Research Record | 2020
|