Volume XLII-1/W1
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-1/W1, 129-134, 2017
https://doi.org/10.5194/isprs-archives-XLII-1-W1-129-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-1/W1, 129-134, 2017
https://doi.org/10.5194/isprs-archives-XLII-1-W1-129-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

  31 May 2017

31 May 2017

LANE LEVEL LOCALIZATION; USING IMAGES AND HD MAPS TO MITIGATE THE LATERAL ERROR

S. Hosseinyalamdary and M. Peter S. Hosseinyalamdary and M. Peter
  • University of Twente, Faculty of ITC, Enschede, the Netherlands

Keywords: Lane Level Localization, Road Boundaries, Map Matching, Lane Matching

Abstract. In urban canyon where the GNSS signals are blocked by buildings, the accuracy of measured position significantly deteriorates. GIS databases have been frequently utilized to improve the accuracy of measured position using map matching approaches. In map matching, the measured position is projected to the road links (centerlines) in this approach and the lateral error of measured position is reduced.

By the advancement in data acquision approaches, high definition maps which contain extra information, such as road lanes are generated. These road lanes can be utilized to mitigate the positional error and improve the accuracy in position.

In this paper, the image content of a camera mounted on the platform is utilized to detect the road boundaries in the image. We apply color masks to detect the road marks, apply the Hough transform to fit lines to the left and right road boundaries, find the corresponding road segment in GIS database, estimate the homography transformation between the global and image coordinates of the road boundaries, and estimate the camera pose with respect to the global coordinate system.

The proposed approach is evaluated on a benchmark. The position is measured by a smartphone’s GPS receiver, images are taken from smartphone’s camera and the ground truth is provided by using Real-Time Kinematic (RTK) technique. Results show the proposed approach significantly improves the accuracy of measured GPS position. The error in measured GPS position with average and standard deviation of 11.323 and 11.418 meters is reduced to the error in estimated postion with average and standard deviation of 6.725 and 5.899 meters.