Enhancement of SPaT-messages with machine learning based time-to-green predictions

Abstract

The involvement of digital technology has changed the transportation domain significantly in the last decade. The availability of several new data sources (i.e., sensor technology or vehicle technology) postulates for data-driven methodologies that can be incorporated into well-established traffic management systems on a macro- and micro-scopic level. Furthermore, the upcoming developments,such as Vehicle-to-Infrastructure (V2I), open the door for new approaches that allow considering communication between vehicles and infrastructure. Recent evolution in traffic signal control of urban intersections (e.g., actuated signal control, self-control algorithms, etc.) influence the signal phases and result in variable green, red and cycle times. Hence, speed advisory systems would benefit from the information about when the next green phase starts so that vehicles do not have to stop when crossing an intersection. Nevertheless, predictions for residual times of these quantities are not trivial and require a sophisticated modeling approach.