Estimation of Curves and Grades for Planning and Safety Analysis from GIS And GPS

Abstract

Vertical profile and horizontal alignment of the road are used by the Highway Safety Manual to estimate the effect of the of grades and curves on the number of predicted crashes. Nevertheless, not always this resource is available to planners. In addition, large scale planning applications may render the conventional extraction of geometry parameters unfeasible. In this context, this paper proposes a general-purpose algorithm based on both GPS or Geographic Information System databases to extract geometry data from alignments and profiles for use in safety and operation analysis. In special, the horizontal curve component was designed to identify spiral transitions. A genetic algorithm was coded to calibrate the proposed curve algorithm, so that curve radii, length and spiral transitions match real data. A testing two-lane highway facility was selected to evaluate accuracy of the proposed method. Resulting curve parameters estimated by the proposed algorithm were compared to as-built drawings obtained from 26 curves. All curves were identified by the algorithm, and all transitions correctly classified. The estimated radii values were similar to the real ones, which indicates that the proposed method could be validated and applicable to different locations for planning level analysis.