Anticipatory Risk Assessment

Future driver assistance systems focus on traffic safety for all road users. This requires enhanced anticipatory risk assessment methods to determine a measure for the endangerment of all individuals caused by the present traffic situation. The target of the subproject is to develop such methods according to the following steps:

Based on a detailed and accurate environment description of the host-vehicle, which is obtained by its environmental onboard sensors, probable trajectories of all road users close to the host-vehicle are predicted up to several seconds into the future. This prediction is based on both, the current dynamic of the observed objects and digital maps providing information about the tracks of the road.

Furthermore, behavioural models of standard drivers are believed to improve the prediction performance. Therefore, we will investigate and model the typical behaviour of drivers in accident-prone scenarios, such as at intersections, in order to enhance the trajectory prediction.

All possible trajectories are iteratively combined which results in possible future situations of different likelihood. The main part of the project is to develop a metric to assess the risk of these future situations. Moreover, the possible benefit of vehicle-to-vehicle communication as additional sensor information will be researched.

The result of this work can be used by a HMI component to warn the driver or draw the driver's attention to possibly dangerous traffic participants.