Christina Gackstatter

Fusion of clothoid segments for a more accurate and updated prediction of the road geometry

A basic step in the development of advanced driver assistance systems is the perception and interpretation of information on the vehicle environment. In many cases, the upcoming road geometry in front the own vehicle is of particular interest. Although a prediction of the course of the road can be provided based on map data, the accuracy is not sufficient for most driver assistance systems. The goal of this work is to achieve a more accurate and updated prediction of the road geometry at close range. Therefore, the map data is combined with the information from an existing vision-based lane detection system. Since the shape of detected road lines as well as the map data are represented by clothoid curves, the emphasis of this work is placed on the calculation, combination, and connection of clothoid segments.

Stable Road Lane Model Based on Clothoids

In the following, the main concepts of a road lane model that keeps track of an arbitrary number of lane borders are presented. Information from an existing lane detection device, a gyrosensor, and map data are merged and filtered to create a road model with a desired number of road lines. The model is based on clothoids and continuously provides positions, angles, and curvatures of the border lines of the vehicle’s own lane as well as of several neighboring lanes. Particularly on urban roads, in situations with upcoming turning lanes, or when the lane detection system fails to detect road lines, the model can still provide plausible information. This information significantly simplifies the situation analysis in further algorithms that rely on a lane detection system and require detailed information on current road lanes.