GLOSA System with Uncertain Green and Red Signal Phases

Abstract

A discrete-time stochastic optimal control problem was recently proposed to address the GLOSA (Green Light Optimal Speed Advisory) problem in cases where the next signal switching time is decided in real-time and is therefore uncertain in advance. However, there was an assumption that the traffic signal is initially red and turns to green, which means that only half traffic light cycle was considered. In this work, the aforementioned problem is extended considering a full traffic light cycle, consisting of four phases: a certain green phase, during which the vehicle can freely pass; an uncertain green phase, in which there is a probability that the traffic light will extend its duration or turn to red at any time; a certain red phase during which the vehicle cannot pass; and an uncertain red phase, in which there is a probability that the red signal may be extended or turn to green at any time. It is demonstrated, based on preliminary results, that the proposed SDP (Stochastic Dynamic Programming) approach achieves better average performance, in terms of fuel consumption, compared to the IDM (Intelligent Driver Model), which emulates human-driving behavior.