Search

The safety assessment of driving situations is of central importance for collision warning systems, driver assistance systems or autonomous driving vehicles. This article introduces a new method for safely evaluating complex driving situations. Some of the correlations about vehicle movements are described by stochastic models. The missing knowledge is generated by an artificial intelligence method, so that a complete stochastic model is available for safety assessment. This model provides information about cur-rent driving situations such as the probability of a collision with a front vehicle or the proba-bility of a serious collision. From these determined quantities, hazard warnings can be gener-ated, or braking maneuvers can be initiated, but also new driving functions can be analyzed already in the development phase. The application of the method is demonstrated by means of recorded vehicle movement data from a traffic simulation. A direct safety assessment in the vehicle is also possible while driving.