Analysing driver's decision in dilemma zone at signalized intersections under disordered traffic conditions

Delay in the decision-making process of stop or go during the amber phase of the signal cycle often leads to abrupt hard deceleration or red light violations at signalized intersections. The indecisiveness or the dilemma in decision making often results in compromised safety of the road users. The present study attempts to analyze the driver's behaviour in order to make the decision of stop or go and developed a binary logistic regression model while considering different traffic behaviour parameters exhibited and observed after the onset of the amber phase. Empirical vehicular trajectory data from three signalized intersections covering 121 signal cycles and 1347 vehicles are used in the study. The study presents two dilemma zone identification models based on distance from the stop line, focusing on easy-to-use and static driver assistance and dynamicrealtime driver assistance systems. Both the models are observed to show good fit and prediction accuracy. The models are validated internally and externally for their adaptability in the field. The effect of different traffic parameters on the dilemma zone is explored, and a possible real-time application of the dynamic model as a driver assistance system in decision-making is explored.