Accuracy and robustness of road observers with uncertainties for reconstruction of the road elevation profile

Abstract

For intelligent or autonomous vehicles the knowledge of the street course and the road condition are indispensable in order to react as early as possible in case of danger. We want to investigate new possibilities for the observation of the road elevation profile in order to warn the driver or road user e.g. of upcoming large potholes or bad roads. This work is a contribution to develop so-called road observers to determine the road profile directly by the vehicle. The derived road observers considering model uncertainties are compared in order to analyze the potential for the road profile reconstruction in simulation and on a measured reference road. It will be shown that the different road observers can be used to estimate the road elevation profile. One of them (in this work is called as “Kalman-4dof”), which is based on the simplified vehicle model, is less sensitive to parameter variations, so that this observer is going to prefer in further applications.