Title:A Driving Risk Surrogate and Its Application in Car-Following Scenario at Expressway

Abstract:Traffic safety is important in reducing death and building a harmonious society. In addition to studies of accident incidences, the perception of driving risk is significant in guiding the implementation of appropriate driving countermeasures. Risk assessment can be conducted in real-time for traffic safety due to the rapid development of communication technology and computing capabilities. This paper aims at the problems of difficult calibration and inconsistent thresholds in the existing risk assessment methods. It proposes a risk assessment model based on the potential field to quantify the driving risk of vehicles. Firstly, virtual energy is proposed as an attribute considering vehicle sizes and velocity. Secondly, the driving risk surrogate(DRS) is proposed based on potential field theory to describe the risk degree of vehicles. Risk factors are quantified by establishing submodels, including an interactive vehicle risk surrogate, a restrictions risk surrogate, and a speed risk surrogate. To unify the risk threshold, acceleration for implementation guidance is derived from the risk field strength. Finally, a naturalistic driving dataset in Nanjing, China, is selected, and 3063 pairs of following naturalistic trajectories are screened out. Based on that, the proposed model and other models use for comparisons are calibrated through the improved particle optimization algorithm. Simulations prove that the proposed model performs better than other algorithms in risk perception and response, car-following trajectory, and velocity estimation. In addition, the proposed model exhibits better car-following ability than existing car-following models.