Probabilistic Approach to Controlling Dilemma Occurrence at Signalized Intersections

This paper proposes a new probabilistic method of protecting drivers against Type I dilemmas that may occur on the display of a yellow traffic signal. At the onset of yellow, an approaching vehicle may be too close to stop but also too far away to pass the stop line before the end of yellow. The occurrence of this dilemma is affected by factors such as driver preference and perception-reaction time, which cannot be fully determined in real time. In this study, the dilemma likelihood function is derived and a method for the identification of the optimal green extension that minimizes this likelihood is proposed. A property of the likelihood function is noted and is used to devise an efficient search algorithm for the optimal green extension. In selecting an optimal extension time, the positions of all vehicles on the approaches, the vehicle types, the speeds, and even the pavement conditions can be considered. To evaluate the effectiveness of this method, numerical simulations are performed on the basis of data collected in the field. The method was found to reduce the frequency of Type I dilemma occurrence significantly while the average green extension was quite short.