The vehicle’s risk of accident is higher in bend than straight line driving. The statistics showed that 40% of fatal accidents occur in bend configuration. These accidents are still much more serious for a collision. This type of accident is obtained when a vehicle leaves its driven space to find itself in the opposite traffic lane. Note that it is not rare to see vehicles especially cutting the bend road axis at high speed. Therefore, the presence of vehicle in the opposite traffic lane is measured by the residence time. This time can be relatively long according to the type of driving behavior.

The objective of this study is to evaluate the probability of collision with the vehicles circulating in opposite direction. This probability can increase quickly according to the residence time and daily traffic flow. Indeed, we already have realistic models of vehicles trajectories in bend. From which, it will be able to estimate an average residence time of the vehicles in the opposed traffic lane per unit of time. Then, by choosing realistic models of traffic flow (for example a Poisson’s model), the probabilities of collision will be estimated for a given bend.

The experiments have shown the relevance and effectiveness of this method. This result will be able to contribute to the bend dangerousness and behaviours analysis.