With urbanization and globalization increasing overcrowding in our traffic systems, vehicles must increase efficiency, safety and environmentally friendliness. Autonomous vehicle technology is set to play a major role is confronting these challenges. However, the elimination of the driver poses numerous challenges; above all that human assessment and experience must now to be regulated by software and hardware only. Programmed algorithms have to make difficult and controversial decisions about the actions of the vehicle, most extremely, unavoidable collisions in which people are involved.
The aim of the research project is to carry out a technical-ethical evaluation of autonomous driving functions in order to answer the question: is it possible to integrate ethical behavior into path and behavior planning of an autonomous vehicle? We will use state of the art algorithms for path and behavior planning and enhance them with ethical theories, in the process translating ethical theories into software. Afterwards, the newly developed algorithms will be analyzed in both well-known and newly defined driving scenarios on a hardware-in-the-loop simulator.
A five-step ethical decision-making model for self-driving vehicles: Which (ethical) theories could guide the process and what values need further investigation?
An ethical trajectory planning algorithm for autonomous vehicles
Autonomous Driving Ethics: from Trolley Problem to Ethics of Risk
AI and Autonomous Driving: Key ethical considerations
AI4People: Ethical Guidelines for the Automotiv Sector–Fundamental Requirements and Practical Recommendations
Maximilian Geißlinger and Franziska Poszler starred in the documentary “Autonomes Fahren: Wie selbstfahrende Autos bei einem Unfall entscheiden“ at Bayerischer Rundfunk – you can watch the video here
The Swiss broadcasting company SRF filmed Maximilian Geisslinger and Franziska Poszler to talk about their project and the ethics of autonomous vehicles. The documentary is published in the program of ‘Einstein’ and can be streamed here