Mining GPS data to augment road models
KDD '99 Proceedings of the fifth ACM SIGKDD international conference on Knowledge discovery and data mining
Route planning and map inference with global positioning traces
Computer Science in Perspective
From GPS traces to a routable road map
Proceedings of the 17th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems
Vlfeat: an open and portable library of computer vision algorithms
Proceedings of the international conference on Multimedia
Hi-index | 0.00 |
In this paper we present a novel algorithm for constructing 3D vector maps for off road navigation using GPS track recordings. The algorithm treats the GPS track recordings as a point cloud and performs track clustering by iteratively aggregating the point cloud. The aggregation is done by simulating attraction forces between samples forming groups of points that belong to the same trail. Next, it constructs clusters centers out of the groups and forms center-lines that connect the cluster centers to create a 3D graph representation. Graph enrichments such as junction clarification are also applied. We demonstrate the results of our algorithm on a simulated dataset of ten tracks for verification purposes and on a dataset of actual GPS tracks of mountain bike trails. An analysis of the geodetic accuracy of the algorithm solution is given using ground control points which were measured in the field.