Positioning in Urban Area by GPS and Monocular Vision Sensor

J.H. Lim, K.H. Choi, H.S. Kim, J.Y. Lee, H.K. Lee

Abstract:	As widely known, GPS (Global Positioning System) cannot provide the position solutions continuously in urban area due to insufficient number of visible satellites. In order to overcome the shortcoming of GPS, hybrid positioning methods have been extensively studied to complement GPS with various other aiding sensors. To bound positioning errors in urban area effectively, this paper proposes a novel hybrid positioning method combining GPS and monocular vision sensor. Based on the rectangular geometries of roads appearing in urban area, it can be assumed that vehicles usually move in straight direction except few cases for right/left turn and lane change. By comparing the two detected vanishing points from two consecutive image frames, it can be checked whether the vehicle is moving straight or not. Once the straight movement condition is confirmed, which occurs mostly during the drive in urban area, the incremental position can be extracted effectively for the coasting of vehicle’s position even when the number of visible satellites is insufficient. To detect vanishing points in real-time with feasible computational burden, the proposed method combines Hough-Transformation, ROI(Region Of Interest), and Color separation methods. As compared with conventional vision-aided GPS methods, the proposed method does not require any external map information and multiple vision sensors to extract depth information. By an experiment result utilizing field-collected measurements, feasibility of the proposed method is demonstrated.
Published in:	Proceedings of the ION 2013 Pacific PNT Meeting April 23 - 25, 2013 Marriott Waikiki Beach Resort & Spa Honolulu, Hawaii
Pages:	388 - 393
Cite this article:	Lim, J.H., Choi, K.H., Kim, H.S., Lee, J.Y., Lee, H.K., "Positioning in Urban Area by GPS and Monocular Vision Sensor," Proceedings of the ION 2013 Pacific PNT Meeting, Honolulu, Hawaii, April 2013, pp. 388-393.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In