Comparison of Two Simulation Approaches to Safety Assessment: Cellular Automata and SSAM

A study is conducted to compare two simulation methods for estimating conflicts between road users. An improved cellular automata (CA) model is proposed to estimate the occurrences and severity of traffic conflicts (both vehicle-vehicle and vehicle-pedestrian) at signalized intersections. The proposed CA model is compared with a calibrated method of a surrogate safety assessment model (SSAM) based on Vissim. Simulated conflicts from both methods are compared with observed vehicle conflicts from automated vehicle tracking for both occurrences and severity. Simulation results show that the CA approach is able to replicate realistic conflicts. However, SSAM tends to overestimate occurrences and underestimate the severity of rear-end and lane-change conflicts. SSAM has also been found to overestimate the severity of crossing conflicts. Furthermore, the proposed CA model is able to estimate conflicts between vehicles and pedestrians.