• Organisation

The prevention of risky situations is one of the main tasks in autonomous driving (AD) and intelligent driving assistant systems (ADAS). Uncertainty in the traffic participants’ behavior and the sensor measurements leads to critical situations, which have to be anticipated by appropriate risk prediction approaches. The risk prediction itself requires dedicated driver models which are interaction sensitive and computationally cheap, to efficiently simulate how a scene might evolve. In this paper, we present a new driver model which is aware of the usual risks encountered in normal driving scenarios. It can cope with longitudinal as well as lateral collision risks, and adjusts its behavior by minimizing the expected integral risk. We show how our model is suited for coping with parallel lane scenarios like overtaking, following and in-between positioning by analyzing its behavior and stability.

Keywords: Risk Assessment, Safety, Trajectory Prediction, Trajectory Planning