Behavioural Modelling of Automated Driving Logic for Advanced Driver Assistance Systems at Signalised Intersections Using Microsimulation

Abstract

The rapid advancement of Advanced Driver Assistance Systems (ADAS) and Automated Vehicles (AVs) has increased global interest in understanding how automated driving logic influences traffic performance, safety, and environmental outputs. While most existing ADAS rely on kinetic thresholds such as headway or speed, empirical evidence indicates that drivers often adopt their behaviours (both positive and negative) when interacting with automated technologies. This study investigates the impact of different automated driving behavioural logics (Normal, Cautious, and Aggressive) on the traffic performance at an urban signalised intersection in the UAE through integrating behavioural adaptation theory with agent-based microsimulation using PTV VISSIM Software. A calibrated and validated simulation model based on real-word traffic data was developed to evaluate seven scenarios representing different levels of automated vehicle penetration and behavioural compositions. Automated driving logic was implemented using behavioural parameters derived from the CoEXit automated vehicle framework.The results indicate that the scenario with 100% normal automated vehicles and a high-penetration mixed AV configuration yields modest improvements in delay, queue length, and traffic flow stability. In contrast, caution and aggressive automated driving increase delay, emissions, and stop frequency. The findings highlight that uncoordinated automated vehicles penetration may not automatically yield improvements in traffic performance, safety, or environment under mixed-traffic conditions. Integrating behavioural modelling into ADAS and automated vehicle development is therefore essential for achieving safer and more efficient transitions towards automated mobility.