Advisory versus automated dynamic eco-driving at signalized intersections: lessons learnt from empirical evidence and simulation experiments

Abstract

Research in the field of dynamic eco-driving has been primarily coupled with connected and automated vehicles which are equipped with automation functions that can accurately execute energy-efficient speed advice. Advisory dynamic eco-driving that entails driver adaptation to energy-efficient speed advice has received lesser attention although mixed traffic is expected to prevail in the forthcoming decades. This study developed a decision tree model based on real-world data collected during the pilot deployment of an advisory speed advice service along an urban arterial corridor to emulate driver adaptation to speed advice. The decision tree model was integrated into the control logic of an enhanced velocity planning algorithm to replicate the behavior of manually driven connected vehicles along dynamic eco-driving service zones in a microscopic traffic simulation environment. The conducted simulation analysis encompassed scenarios with varying penetration rates of advisory dynamic eco-driving technology, automated dynamic eco-driving technology and manually driven unequipped vehicles. Evaluation of simulation scenarios was based on the estimation of several environmental, traffic efficiency and surrogate safety measures. Simulation results indicated that performance of advisory dynamic eco-driving depends on driver adaptation to speed advice and ranges between that of manually driven unequipped vehicles and its automated counterpart. Moreover, geometrical and operational characteristics of intersection approaches comprising dynamic eco-driving service zones can influence driver adaptation to speed advice. Environmental, safety and traffic efficiency benefits are maximized in the case of vehicle fleets fully equipped with automated dynamic eco-driving systems.