Eco-driving at Signalized Intersections: What is Possible in the Real-World?

A significant amount of fuel is wasted at signalized intersections due to unnecessary braking and idling. Connected Vehicle technologies make future signal phasing and timing (SPaT) information available to drivers. Recent research which utilized SPaT on eco-driving has verified its potential to reduce fuel consumption in simulations and controlled lab tests. In this paper, we propose an eco-driving algorithm and use real world driving data of 609 human-driven trips undertaken in Ann Arbor, MI. The proposed eco-driving method shows potential fuel savings of 40-50% while matching human travel time. Results were formulated under the assumption that the vehicle is operating in free-flow traffic, utilizing SPaT of the two consecutive signalized intersections, and equipped with a continuously variable transmission and an internal combustion engine. For each human-driven trip recorded, the proposed method uses Dynamic Programming to determine globally optimal trajectories of three different eco-driving policies: fuel optimal policy, time-optimal policy, and balanced eco-driving policy. Given the same initial and the final conditions as those of human-driven trips, comparisons are made to demonstrate the real-world benefits of the eco-driving policies. These results can serve as an upper bound of fuel and travel time saving potential of the eco-driving in the vicinity of signalized intersections.