Vehicle dynamic risk assessment at intersections based on gravitational model

Intersections, as the "bottleneck" of the road network, involve a large amount of traffic flow and conflicts, posing high accident risks. Accurate risk assessments of vehicles within intersections can effectively reduce accident rates and improve intersection safety. Traditional vehicle dynamic risk assessment methods often establish models from both dynamic and static factors that affect vehicle operation. These methods have complex calculation processes, and mainly use relative speed to represent the impact of object speed on risk, which cannot accurately describe the impact of different movement trends on vehicle driving risk. This study proposes a dynamic risk assessment method for intersection vehicle operation based on the gravitational model. This model uses pseudo-velocity to characterize the impact on the driving risk of the velocity of environmental elements to the ego vehicle, and then employs the gravitational model to uniformly quantify the potential risk posed by different elements to the ego vehicle. The rationality of the model is verified through numerical simulations of typical car-following scenarios. It’s effectiveness in risk warning is then demonstrated by comparing its capabilities with the Time to Collision (TTC) method in a typical car-following scenario. The applicability of the model to typical conflict scenarios at intersections is verified through the risk assessment of vehicle merging conflict scenarios and typical left-turn merging conflict scenarios at an intersection in Suzhou, China. This model can provide a basis for vehicle warning systems, prevent traffic accidents at intersections, and enhance intersection safety.