The development of autonomous vehicles faces not only technical obstacles, but also ethical questions that must be addressed. Autonomous vehicles must be able to handle unexpected situations and make decisions in the event of an impending accident.
To address these ethical questions, researchers at TUM have developed the first ethical algorithm to fairly distribute risk. The algorithm considers various risks and makes ethical choices from among thousands of possible behaviors in a fraction of a second.
The basic ethical parameters that the software’s risk evaluation is based on were defined by an expert panel as a written recommendation on behalf of the EU Commission in 2020. The recommendation includes basic principles such as prioritizing the worst-off and the fair distribution of risk among all road users. To translate these rules into mathematical calculations, the research team classified vehicles and people based on the risk they pose to others and their willingness to take risks. The algorithm was then instructed not to exceed a maximum acceptable risk in various road situations and factors in responsibility, such as obeying traffic regulations.
Previous approaches only considered a small number of possible maneuvers in critical situations and resulted in the vehicle stopping in unclear cases. The new algorithm results in more possible degrees of freedom with less risk for all participants. For example, in a scenario where an autonomous vehicle wants to overtake a bicycle while a truck is approaching in the opposite lane, the algorithm evaluates the risk posed to each vehicle and makes a decision based on the acceptable level of risk. Aggressive maneuvers are avoided, and the vehicle does not freeze and abruptly stop.
The researchers emphasized that even algorithms based on risk ethics cannot guarantee accident-free road traffic. In the future, cultural differences in ethical decision-making must also be considered. The algorithm developed at TUM has been validated in simulations and will soon be tested on the street using the research vehicle EDGAR. The code is available as open-source software, and TUM is contributing to the development of safe autonomous vehicles.