Effects of a Probability-Based Green Light Optimized Speed Advisory on Dilemma Zone Exposure 2020-01-0116
Green Light Optimized Speed Advisory (GLOSA) systems have the objective of providing a recommended speed to arrive at a traffic signal during the green phase of the cycle. GLOSA has been shown to decrease travel time, fuel consumption, and carbon emissions; simultaneously, it has been demonstrated to increase driver and passenger comfort. Few studies have been conducted using historical cycle-by-cycle phase probabilities to assess the performance of a speed advisory capable of recommending a speed for various traffic signal operating modes (fixed-time, semi-actuated, and fully-actuated). In this study, a GLOSA system based on phase probability is proposed. The probability is calculated prior to each trip from a previous week’s, same time-of-day (TOD) and day-of-week (DOW) period, traffic signal controller high-resolution event data. By utilizing this advisory method, real-time communications from the vehicle to infrastructure (V2I) become unnecessary, eliminating data-loss related issues. The effects of three different advice approaches (conservative, balanced, and aggressive) on dilemma zone exposure are analyzed. Proof of concept is carried out by virtually driving through a test-route composed of an arterial that had historical high-resolution traffic signal event logs for a series of actuated-coordinated traffic signals during different TOD and DOW. A comparison was performed between unadvised and GLOSA advised trips obtained from approximately 486000 simulated trajectories. Results were obtained by analyzing the vehicle’s probability of stopping from utilizing Traffic Engineering dilemma zone theory. Reductions of 93% in the amount of hard brakings and 96% in the number of crossings through red light were observed with the proposed system. This data suggests the feasibility of a probability-based advisory, as well as the viability of utilizing the proposed GLOSA system to minimize dilemma zone exposure.
Enrique Saldivar-Carranza, Howell Li, Woosung Kim, Jijo Mathew, Darcy Bullock, James Sturdevant
Purdue University, Indiana Department of Transportation