Learning-Based Cooperative Adaptive Cruise Control

Traffic congestion and the occurrence of traffic accidents are problems that can be mitigated by applying cooperative adaptive cruise control (CACC). In this work, we used deep reinforcement learning for CACC and assessed its potential to outperform model-based methods. The trade-off between distanc...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Jonas Mirwald, Johannes Ultsch, Ricardo de Castro, Jonathan Brembeck
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	cooperative adaptive cruise control C2C communication deep reinforcement learning Modelica vehicle modeling Materials of engineering and construction. Mechanics of materials TA401-492 Production of electric energy or power. Powerplants. Central stations TK1001-1841
Acceso en línea:	https://doaj.org/article/d637cd7a680a413f9746628cfcd1b366
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:d637cd7a680a413f9746628cfcd1b366
record_format	dspace
spelling	oai:doaj.org-article:d637cd7a680a413f9746628cfcd1b3662021-11-25T15:56:49ZLearning-Based Cooperative Adaptive Cruise Control10.3390/act101102862076-0825https://doaj.org/article/d637cd7a680a413f9746628cfcd1b3662021-10-01T00:00:00Zhttps://www.mdpi.com/2076-0825/10/11/286https://doaj.org/toc/2076-0825Traffic congestion and the occurrence of traffic accidents are problems that can be mitigated by applying cooperative adaptive cruise control (CACC). In this work, we used deep reinforcement learning for CACC and assessed its potential to outperform model-based methods. The trade-off between distance-error minimization and energy consumption minimization whilst still ensuring operational safety was investigated. Alongside a string stability condition, robustness against burst errors in communication also was incorporated, and the effect of preview information was assessed. The controllers were trained using the proximal policy optimization algorithm. A validation by comparison with a model-based controller was performed. The performance of the trained controllers was verified with respect to the mean energy consumption and the root mean squared distance error. In our evaluation scenarios, the learning-based controllers reduced energy consumption in comparison to the model-based controller by 17.9% on average.Jonas MirwaldJohannes UltschRicardo de CastroJonathan BrembeckMDPI AGarticlecooperative adaptive cruise controlC2C communicationdeep reinforcement learningModelica vehicle modelingMaterials of engineering and construction. Mechanics of materialsTA401-492Production of electric energy or power. Powerplants. Central stationsTK1001-1841ENActuators, Vol 10, Iss 286, p 286 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	cooperative adaptive cruise control C2C communication deep reinforcement learning Modelica vehicle modeling Materials of engineering and construction. Mechanics of materials TA401-492 Production of electric energy or power. Powerplants. Central stations TK1001-1841
spellingShingle	cooperative adaptive cruise control C2C communication deep reinforcement learning Modelica vehicle modeling Materials of engineering and construction. Mechanics of materials TA401-492 Production of electric energy or power. Powerplants. Central stations TK1001-1841 Jonas Mirwald Johannes Ultsch Ricardo de Castro Jonathan Brembeck Learning-Based Cooperative Adaptive Cruise Control
description	Traffic congestion and the occurrence of traffic accidents are problems that can be mitigated by applying cooperative adaptive cruise control (CACC). In this work, we used deep reinforcement learning for CACC and assessed its potential to outperform model-based methods. The trade-off between distance-error minimization and energy consumption minimization whilst still ensuring operational safety was investigated. Alongside a string stability condition, robustness against burst errors in communication also was incorporated, and the effect of preview information was assessed. The controllers were trained using the proximal policy optimization algorithm. A validation by comparison with a model-based controller was performed. The performance of the trained controllers was verified with respect to the mean energy consumption and the root mean squared distance error. In our evaluation scenarios, the learning-based controllers reduced energy consumption in comparison to the model-based controller by 17.9% on average.
format	article
author	Jonas Mirwald Johannes Ultsch Ricardo de Castro Jonathan Brembeck
author_facet	Jonas Mirwald Johannes Ultsch Ricardo de Castro Jonathan Brembeck
author_sort	Jonas Mirwald
title	Learning-Based Cooperative Adaptive Cruise Control
title_short	Learning-Based Cooperative Adaptive Cruise Control
title_full	Learning-Based Cooperative Adaptive Cruise Control
title_fullStr	Learning-Based Cooperative Adaptive Cruise Control
title_full_unstemmed	Learning-Based Cooperative Adaptive Cruise Control
title_sort	learning-based cooperative adaptive cruise control
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/d637cd7a680a413f9746628cfcd1b366
work_keys_str_mv	AT jonasmirwald learningbasedcooperativeadaptivecruisecontrol AT johannesultsch learningbasedcooperativeadaptivecruisecontrol AT ricardodecastro learningbasedcooperativeadaptivecruisecontrol AT jonathanbrembeck learningbasedcooperativeadaptivecruisecontrol
_version_	1718413408989085696

Learning-Based Cooperative Adaptive Cruise Control

Ejemplares similares