Development of a Surrogate Conflict Indicator for Freeway Exits Using Trajectory Data

Crash occurrences are random rare events which are difficult to detect and reconstruct. Evaluating traffic safety using crash surrogate measures instead of historical crash data is attracting more and more attention. The trajectory, a continuous function of a vehicle's temporal-spatial kinestate, is capable of depicting the process of crash potentials by means of theoretical analysis or simulation. The trajectory provides easy access to accurate calculation of the commonly used safety surrogate indicators like TTC and PET, which are difficult to obtain in practice. TTC and PET are the time differences between two vehicles during a quasi accident process. Compared with the situation in the final conflicting point, two vehicles may encounter a smaller distance or a bigger speed variation while approaching the conflict point. TTC and PET are deficient in describing abreast driving, or distinguishing the severity levels for approximately equal TTC (PET) cases. To remedy these shortcomings, this paper proposes a surrogate indicator Kj, the ratio of conflicting distance divided by relative speed. And an exponential model is developed to predict the conflict probability. Kj and the conflict probability are both time frame based, illustrating the changing process at each time frame during a conflict phase. The indicator Kj and conflict probability make it easier to describe the conflict mechanism and distinguish the levels of conflict severity.