This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Machine Learning-Based Eco-Approach and Departure: Real-Time Trajectory Optimization at Connected Signalized Intersections

Journal Article
13-03-01-0004
ISSN: 2640-642X, e-ISSN: 2640-6438
Published October 13, 2021 by SAE International in United States
Machine Learning-Based Eco-Approach and Departure: Real-Time Trajectory Optimization at Connected Signalized Intersections
Sector:
Citation: Esaid, D., Hao, P., Wu, G., Ye, F. et al., "Machine Learning-Based Eco-Approach and Departure: Real-Time Trajectory Optimization at Connected Signalized Intersections," SAE J. STEEP 3(1):2022, https://doi.org/10.4271/13-03-01-0004.
Language: English

Abstract:

Taking advantage of communication and sensing technology, trajectory optimization at signalized intersection (i.e., Eco-Approach and Departure) based on information from vehicle-to-infrastructure (V2I) communications has been proven to be effective to improve vehicle energy efficiency while guaranteeing safety and mobility. However, existing approaches are either rule-based models or optimization models, which cannot achieve optimality and computational efficiency at the same time. In this article, we propose and test a novel learning-based approach, machine learning trajectory planning algorithm (MLTPA), to achieve real-time optimization by training a machine learning model to approximate the solution from a previously developed optimization-based method named graph-based trajectory planning algorithm (GBTPA). Five types of machine learning techniques, including linear regression, k-nearest neighbors, decision tree, random forest, and multilayer perceptron (MLP) neural network, are compared in terms of the prediction accuracy, and the random forest method is finally selected. The proposed MLTPA reduces computation time from tens of seconds to a few milliseconds. Simulation results illustrate that MLTPA can achieve median 5.0%-6.20% improvement on energy savings over multiple simulation runs. The proposed method also has the potential to approximate other trajectory planning algorithms to achieve real-time performance while ensuring optimality.