Measurement uncertainty determination and curve-fitting algorithms for development of accurate digital maps for advanced driver assistance systems

Digital maps can provide support for numerous advanced driver assistance systems (ADAS) aimed at improving road safety. These new uses require more highly detailed and precise maps. The use of a datalog vehicle to collect roadway geometry data can fulfil these specifications. This paper presents fast, accurate measurement with an on-board inertial system together with a method to evaluate measurement uncertainty, particularly for any variables obtained indirectly. It also presents an algorithm for segmentation and fitting geometric curves to the experimental points, following current highway design standards. The algorithms have been applied to real road measurements. Segmentation has been done in straight alignments, circular curves and transition curves whose characteristic parameters are calculated. It has been seen that with a very small data set it is possible to reconstruct the measured geometry with few discrepancies regarding the experimental points.