Through modern driver assistant systems, algorithmic decisions already have a significant impact on the behavior of vehicles in everyday traffic. This will become even more prominent in the near future considering the development of autonomous driving functionality. The need to consider ethical principles in the design of such systems is generally acknowledged. However, scope, principles and strategies for their implementations are not yet clear. Most of the current discussions concentrate on situations of unavoidable crashes in which the life of human (...) beings is existentially affected. In this paper, we argue that ethical considerations should be mandatory for any algorithmic decision of autonomous vehicles, instead of a limitation to hazard situations. Such an ethically aligned behavior is relevant because autonomous vehicles, like any other traffic participants, operate in a shared public space, where every behavioral decision impacts the operational possibilities of others. These possibilities concern the fulfillment of a road-user’s safety, utility and comfort needs. We propose that, to operate ethically in such space, an autonomous vehicle will have to take its behavior decisions according to a just distribution of operational possibilities among all traffic participants. Using an application on a partially-autonomous prototype vehicle, we describe how to apply and implement concepts of distributive justice to the driving environment and demonstrate the impact on its behavior in comparison to an advanced but egoistic decision maker. (shrink)

This work investigates which conditions lead to co-driver discomfort aside from classical motion sickness, what characterizes uncomfortable situations, and why these conditions have a negative effect. The automobile is called a “passenger vehicle” as its main purpose is the transportation of people. However, passengers in the car are rarely considered in research concerning driving discomfort. The few studies in this area focus on driver discomfort, automated vehicles, or driver assistant systems. An earlier public survey indicated that discomfort is also a (...) relevant problem for co-drivers. In this paper, these results are confirmed and extended through an online questionnaire with N = 119 participants and a detailed follow-up interview study with N = 24 participants was conducted. The results of the online questionnaire show that co-driver discomfort is a widespread problem. The interviews indicate that the driving style is one factor contributing to co-driver discomfort, in particular close following or fast driving. In those situations, participants experienced a feeling of being exposed, which additionally contributed to their discomfort. Uncomfortable situations were also perceived as safety critical. A model for possible cognitive origins of discomfort in co-drivers, extending theories from the areas of stress and self-regulation, is developed based on the results. Co-driver discomfort is a common problem, highlighting the relevance of further research on supporting co-drivers. The reported correlations and the proposed model can help to explain the origin of this discomfort. The results provide a foundation for the future design of interventions like human machine interfaces aiming at reducing co-driver discomfort. (shrink)