A Critical Evaluation of Eco-Driving Strategies for Connected Autonomous Electric Vehicles at Signalized Intersections

Abstract

Signalized intersections are significant spots of energy consumption because of frequent stop-and-go behavior. Eco-driving aims to reduce energy usage by optimizing driving behavior. Researchers have reviewed optimization-based method while lack of them reviewed the learning-based approaches. This work critically reviewed two different types of approach. In addition, one well-known rule-based car-following model and two state-of-the-art optimization-based and learning-based methods are selected to test in a signalized intersections environment with the metrics of energy consumption, travelling time and algorithm execution time. The experiment results show that the travelling time of three algorithms are similar, while the energy consumption of the learning-based method and optimization-based method are 30.72% and 51.82% less than that of the rule-based method respectively. However, due to algorithm execution time, the optimization-based method is not suitable to be used in real-time.