Strategic crossing decisions in vehicle interactions at uncontrolled intersections: A networked driving simulator study

At uncontrolled intersections, without traffic signals or explicit right-of-way guidance, drivers interacting from different directions depend on their perception and real-time judgment to make strategic decisions about when and how to cross or yield. This uncertainty in decision-making, if misjudged, may result in vehicle collisions. In this study, we developed a networked two-vehicle driving simulator platform, conducted experiments to simulate the interaction between straight-going (SG) and left-turn (LT) vehicles at uncontrolled intersections, and focused on exploring the contributing factors that may influence drivers’ crossing decisions. By analyzing the varying sequences of vehicle acceleration with changes in distance to conflict (DTC), we first identified the critical transition state which marks the shift from stable to dynamic decision-making, helping determine the moment that drivers make crossing or yielding decisions. Then, by examining vehicle operational indicators at the critical transition state and incorporating human factors, a correlated random parameter random regret minimization binary logit model was developed to investigate the contributing factors influencing crossing decisions made by SG and LT drivers. The results indicate that individual speed and DTC have no significant impact on the crossing decisions of either SG or LT drivers, but variables such as speed difference and DTC difference between the two vehicles do, which highlights the importance of conducting two-vehicle simulation in capturing interaction variables. Furthermore, in terms of human factors, penalty records are found to significantly affect the crossing decisions of both SG and LT drivers. Income level has significant influence on SG crossing decisions, while gender is a significant factor influencing LT crossing decisions. Additionally, this study also reveals that for SG vehicles, the trade-off value between the speed difference and the DTC difference is mostly concentrated between 5 and 6, meaning that each 1 m/s speed difference corresponds to an impact of 5 to 6 m of DTC difference on crossing decisions. For LT vehicles, the trade-off values were primarily between 3 and 5. These findings would be valuable for developing connected vehicle information systems and for supporting policy-making efforts to reduce crash risk.