2017 Volume 3 Issue 2 Pages A_44-A_52
In this study, we developed a method to automatically correct positional errors of digital road map using probe data. Some existing methods minimize a sum of squares of the distances between the positions derived from probe data and links in digital road map database, but they may fail to update nodes’ coordinates with a high degree of accuracy because of bias of distribution of the position information in a section of a link, which represents more than two lanes. In order to reduce this effect, we assumed the distribution could be modeled by a probability model using mixed distributions, and developed a method for updating the nodes’ position with estimated model’s parameters. We first considered the optimal calculation unit to update nodes’ coordinates by using the least-square method. After that, we conducted experiments, and it was found that better updating results were obtained than those by using the existing method.