Three Years CoInCar: What Cooperatively Interacting Cars Might Learn from Human Drivers

Abstract The CoInCar project’s overall aim is to contribute to the development of cooperative driving. Cooperatively interacting cars will conduct maneuvers based on information perceived by and shared with other road users as well as the infrastructure. However, drivers might face problems understanding the vehicles’ behavior, which could lead to critical situations. In the last three years, we investigated human behavior in potentially cooperative traffic situations, like in lane change scenarios and at intersections. Results showed that cooperative behavior in lane changing scenarios on highways depends beside other factors on the way the other drivers are indicating their intention to change lanes. In inner-city intersection scenarios, a comparison between an equal narrow passage and a complex T-junction scenario showed that human drivers prefer to drive firstly in simple situations but to give priority to other cooperation partners in complex situations. Furthermore, the high-frequency compound (HFC) of the steering angle could be identified as the relevant driving parameter for intention recognition. Moreover, for further understanding of the decision-making process, we analyzed cooperative situations with the Natural Decision Making approach. In situations in which the behavior of an automated vehicle differs from the human’s expectations, an explanation will be needed. Therefore, knowledge about human behavior in cooperative situations helps understanding human expectations, which will be relevant in the process of designing interfaces for cooperatively interacting vehicles.