Traffic microsimulation for road safety assessments of vehicle automation scenarios: Model comparison and sensitivity analysis

The possible implementation of Automated Vehicles (AVs) in the traffic flow implies that future safety conditions should be assessed by considering promiscuous traffic scenarios (composed of traditional vehicles and AVs). However, real safety performance data related to AVs are still scarce and unreliable. Thus, road safety assessments of future scenarios should necessarily be based on traffic simulations and surrogate safety measures (i.e., related to traffic conflicts, which are different from observed crashes). In the context of traffic simulation, the first step consists in assessing the currently available traffic models to match their characteristics with the possibility of reliably simulating AVs and their behavior in traffic, according to the specific variables of the investigated site (i.e., road type, road geometric design, number of intersections). After the model selection stage, a sensitivity analysis is necessary to identify the most crucial model parameters and their influence on the simulation output, with particular regard to those affecting safety performances. The sensitivity analysis paves the way for making traffic modelers aware of the effects of possible modifications to the simulation parameters in case of AV scenarios. In this research, this procedure was tested on a two-way two-lane rural road network. Among the analyzed traffic models, the Gipps model was highlighted as the most suitable to account for AVs in this specific context. The sensitivity analysis revealed that the most influencing parameters of the Gipps model in the context of safety assessments are: clearance, safety margin factor and sensitivity factor, which were revealed to be greatly influential on all types of traffic conflicts. Two other control tests with different simulated vehicle types were run to highlight the stability of the results coming from the sensitivity analysis in other scenarios.