The systems of autonomous vehicles are designed to avoid collisions under all circumstances, especially with pedestrians. But that doesn’t always work. The approach of the Canadian scientists therefore concentrates on minimizing the damage: if their software recognizes that a collision cannot be avoided, it analyzes all of the available options incredibly quickly and selects the one with the lowest anticipated damage. It uses pre-defined mathematical calculations to do this. Of course, the ethical questions remain unanswered – such as whether the lives of all people involved in an accident should be assumed to have the same value. If general rules are ever established in this area, they can be integrated into the system.

The model of a so-called Vision Zero, a world of traffic without any accident deaths, is therefore not one to which the Canadian researchers are currently subscribing, at least for the foreseeable future. The idea that autonomous driving could prevent all accidents in future is a myth, in their view. Nevertheless, a study by the Association of German Chambers of Industry and Commerce (DIHK) on autonomous driving at least came to the conclusion that “in many people’s view” accidents that are the result of human error could be avoided almost entirely.