Using Analytics in the Implementation of Vertical and Horizontal Curvature in Route Calculation

Determining the best route for logging trucks is difficult as many road features needs to be considered. There are methodologies to determine a good weight setting for these features based on inverse optimization, allowing them to be combined into one objective. The distance of the minimum-cost route on a network where the arc costs are determined by this weighted objective will then form the basis of invoicing. From detailed collected and agreed routes acting as the most preferred solutions, we can evaluate the quality of our proposed approach. Multiple sources of information are used, e.g., lidar, aerial photos, and GIS databases are collected and made available. There has been a demand from end-users to also include vertical and horizontal curvature as road features in the route selection. Since such values are not directly collected or stored in databases, they must be analyzed and estimated using geographical information. We propose a methodology for computing vertical and horizontal curvatures that represent the perception of such features, as well as processes to clean and complement inaccurate coordinates in the data. The proposed decision tools work very well in practice and are implemented in the intelligent transport planning system called calibrated route finder (CRF). CRF has been used on national level for distance calculation and route selection on since 2010, and the proposed use of vertical and horizontal curvatures was implemented during 2014. Today, over 50% of all forest sector transport invoicing in Sweden is based on the system.