Model-based threat assessment in semi-autonomous vehicles with model parameter uncertainties

In this paper, we consider model-based threat assessment methods which rely on vehicle and driver mathematical models and are based on reachability analysis tools and set invariance theory. We focus on the parametric uncertainties of the driver mathematical model and show how these can be accounted for in the threat assessment. The novelty of the proposed methods lies in the inclusion of the driver model uncertainties in the threat assessment problem formulation and in their validation through experimental data. We show how different ways of accounting for the model uncertainties impact the capabilities and the effectiveness of the proposed algorithms in detecting hazardous driving situations.