Model-based identification of road profiles and road roughness indicators using vehicle measurements

We propose and discuss a computational method to derive road profiles and road roughness indicators based on vehicle models and vehicle measurements. Models of simple complexity are used, such as quarter-car and half-car models. The needed measurements require only moderate effort, e.g., vertical accelerations are sufficient. The derivation task is formulated as inverse problem and is numerically solved using the Bayesian inversion approach, which derives conditioned distributions of the unknowns. Thus, it automatically allows to quantify uncertainty. The method enables the computation of road profiles and road roughness indicators on (very) long tracks or for very large databases (measurement campaigns), respectively, in acceptable computation time, thus, it is also suited for on-board applications.