Recognition of dangerous situations within a cooperative group of vehicles

Conference paper


Thomas Batz
Kym Watson
Jürgen Beyerer


Proceedings of IEEE Intelligent Vehicles Symposium, 2009.




IEEE Intelligent Vehicles Symposium, Xi'an, China, June 3 - 5, 2009

We consider the recognition of dangerous situations in vehicle traffic. Unscented Kalman filters are used to predict vehicle trajectories within a short prediction horizon [t0; t0+Δt]. Based on this prediction, for each vehicle pair the mutual distance is computed for [t0; t0+Δt], whereby the distance accounts for the geometric distance, for the prediction uncertainties as well as for the spatial dimensions of the vehicles. If at least one of the mutual distances falls below a distance threshold ε within [t0; t0+Δt], then a dangerous situation arises for the cooperative group and may lead to an autonomous cooperative driving manoeuvre. This approach allows the usage of the system in a mixed environment (only some vehicles are cooperative and cognitive). Obstacles can also be handled. The key issues in this ongoing research work are the recognition and classification of dangerous situations and the formation of a cooperative group constituting an operational unit. A common relevant picture within a group coordinator fuses the necessary information from all cooperative vehicles of the group and forms the basis for situation recognition and classification. This paper is a step to expand a Cooperative Collision Warning System (CCWS) to an integrated Cooperative Collision Avoidance and Cooperative Collision Mitigation System (CCAMS).