Evaluation of automated driving safety in urban mixed traffic environments

Abstract

Conflicting driving behaviours between automated vehicles and manually driven vehicles may compromise driving safety. The aim of this study is to analyse the safety of mixed traffic on urban roads. The driving simulation tests were conducted using a multi-agent driving simulator, which allows real-time synchronization of multiple simulators. These data were further processed to derive the driving behaviour parameters of manually driven vehicles in VISSIM traffic simulations. Driving safety evaluation indicators included conflict-related indicators, as well as individual safety indicators. The safety evaluation indicators were normalized through min–max normalization, and the risk scores were summed to evaluate the urban roads. The analysis revealed that driving safety was poor at unsignalized intersections with a market penetration rate of 10% and 50% and at signalized intersections with traffic islands and a market penetration rate of 100%, where conflicts arise from the deceleration of leading vehicles and lane changes. This finding is about the driving behaviour of automated vehicles, which maintain a greater distance from the leading vehicle than manually driven vehicles, resulting in poorer driving safety due to lane changes rather than deceleration. Using the findings of this study, criteria for assessing the safety of mixed traffic situations in existing road infrastructures can be established.