Abstract
The aim of this project is to develop novel multi-agent traffic simulation software capable of predicting the impact of automated driving systems on traffic safety. The software is designed to incorporate variables for human perception and recognition, decision-making, and errors, dependent on age and gender. In this study, the software was applied to a 6 x 3 km2 area in Tsukuba city. Over a simulated period of time that included more than 500 agents (vehicles, drivers, and pedestrians), the software simulated and identified at least five types of accidents. Different automated driving technology penetration scenarios were applied to conduct preliminary estimations of the potential impact of these technologies on safety.
