Safety evaluation of signalized intersections with automated vehicles at various penetration levels based on conflict analysis of simulated traffic

Automated vehicles (AVs) are expected to offer great benefits by potentially reducing crashes. The safety at signalized intersections is influenced by several factors, one of them being the driving behavior. By introducing AVs on the roads, the unpredictability of this factor will potentially decrease and eventually, reduce crashes. By using microsimulation, it was possible to use simulated traffic conflicts as indicators of potential crashes, to analyze the potential safety of signalized intersections in the presence of automated vehicles. The objective was to compare crash frequency for signalized intersections at various AVs penetration levels (0%, 50% and 100%) by using prediction models that relate crashes to conflicts. Furthermore, the effect on crashes of introducing hypothetical left turn treatments was also evaluated. The results indicated that intersection safety may improve in the presence of AVs. However, the safety effects of treatments may be reduced compared to the effects with no AVs.