Improving coordinated actuated control performance by measuring “early return to green” using probe vehicle data

Coordinated actuated control is designed to respond to fluctuations in traffic demand on non-coordinated phases, allowing unused green time to be reallocated to the coordinated phase. As a result, when non-coordinated phases gap out early, the coordinated green starts earlier, which is known as “Early Return to Green” (ERG). This stochastic variability across cycles adds complexity to the modeling and design of coordination strategies. Traditionally, optimizing corridors with coordinated actuated control has relied on either hard-to-access operational data from signal controllers or sophisticated optimization models that incorporate stochastic signal states. Both approaches pose significant challenges for real-world implementation. In this paper, we introduce a concise and scalable framework for incorporating ERG into the corridor modeling process when designing coordinated actuated plans. Our approach leverages probe vehicle data to estimate the distribution of actual coordination phase green start times. This distribution is then used as weights to construct a weighted probabilistic time-space diagram, which can be used to predict traffic performance in response to signal parameters. In addition, we introduce several metrics to characterize ERG, which can serve as indicators of detector failures. Numerical experiments using the SUMO simulator further validate the proposed method’s accuracy against baselines and demonstrate its robustness at varying probe data penetration rates. A real-world case study demonstrates the practical application of our methodology through cycle length and offset optimization on a seven-intersection corridor under coordinated actuated control. A before-and-after evaluation highlights reductions of up to 27.3% in control delay and 16.3% in corridor travel time during peak hours.