| Titre : |
A digital twin framework for traffic management : urban intersections and highway scenarios |
| Type de document : |
document multimédia |
| Auteurs : |
Amina Aicha Benmechri, Auteur ; Fadoua Amana Hadj Aissa, Auteur ; Mohamed El Amine Ameur, Directeur de thèse |
| Editeur : |
Laghouat : Université Amar Telidji - Département d'informatique |
| Année de publication : |
2026 |
| Importance : |
115 p. |
| Accompagnement : |
1 disque optique numérique (CD-ROM) |
| Note générale : |
Option : Artificial intelligence and data science |
| Langues : |
Anglais (eng) |
| Mots-clés : |
Digital twin Multi-agent reinforcement learning Traffic management SUMO TraCI Python Q-Learning Adaptive Signal Control Highway rerouting Smart cities Intelligent transportation systems |
| Résumé : |
Modern cities have witnessed a rapid escalation in traffic congestion, road accidents, and environmental emissions, driven by continuous urban growth and the increasing number of vehicles on road networks. Despite the pivotal role that intelligent transportation systems play in monitoring and managing traffic, these systems remain fundamentally reactive — responding to events after they occur, and unable to anticipate or predict risks before they worsen. To overcome this limitation, this thesis proposes an integrated framework combining Digital Twin technology and multi-agent reinforcement learning within a three-level closed loop feedback architecture, encompassing: real-time synchronization, predictive simulation, and adaptive recalibration upon detection of statistical data drift.
The proposed framework was implemented using the SUMO (Simulation of Urban MObility) microscopic traffic simulator coupled with the TraCI (Traffic Control Interface) real-time control API for bidirectional communication between the physical simulation and the Digital Twin layer. Two experimental scenarios were designed and evaluated within this environment. The first scenario represents an urban network with two signalized intersections, where Q-learning agents dynamically control signal phases. The second scenario represents a highway equipped with a smart vehicle rerouting system via two alternative routes.
The results yielded substantial improvements compared to the traditional fixedtiming system. In the first scenario, the proposed framework reduced queue length by 60.0%, total delay by 52.9%, and waiting time by 38.0%, while achieving a synchronization accuracy of 76.1%. In the second scenario, the main route queue decreased by 82.8%, average speed increased by 226.2%, CO2 emissions and fuel consumption decreased by 56.3%, and synchronization accuracy reached 95.7%.
These results confirm that the active bidirectional Digital Twin constitutes an effective platform for proactive decision-making, contributing to traffic management that is more efficient, safer, and more sustainable. |
| note de thèses : |
Mémoire de master en informatique |
A digital twin framework for traffic management : urban intersections and highway scenarios [document multimédia] / Amina Aicha Benmechri, Auteur ; Fadoua Amana Hadj Aissa, Auteur ; Mohamed El Amine Ameur, Directeur de thèse . - Laghouat : Université Amar Telidji - Département d'informatique, 2026 . - 115 p. + 1 disque optique numérique (CD-ROM). Option : Artificial intelligence and data science Langues : Anglais ( eng)
| Mots-clés : |
Digital twin Multi-agent reinforcement learning Traffic management SUMO TraCI Python Q-Learning Adaptive Signal Control Highway rerouting Smart cities Intelligent transportation systems |
| Résumé : |
Modern cities have witnessed a rapid escalation in traffic congestion, road accidents, and environmental emissions, driven by continuous urban growth and the increasing number of vehicles on road networks. Despite the pivotal role that intelligent transportation systems play in monitoring and managing traffic, these systems remain fundamentally reactive — responding to events after they occur, and unable to anticipate or predict risks before they worsen. To overcome this limitation, this thesis proposes an integrated framework combining Digital Twin technology and multi-agent reinforcement learning within a three-level closed loop feedback architecture, encompassing: real-time synchronization, predictive simulation, and adaptive recalibration upon detection of statistical data drift.
The proposed framework was implemented using the SUMO (Simulation of Urban MObility) microscopic traffic simulator coupled with the TraCI (Traffic Control Interface) real-time control API for bidirectional communication between the physical simulation and the Digital Twin layer. Two experimental scenarios were designed and evaluated within this environment. The first scenario represents an urban network with two signalized intersections, where Q-learning agents dynamically control signal phases. The second scenario represents a highway equipped with a smart vehicle rerouting system via two alternative routes.
The results yielded substantial improvements compared to the traditional fixedtiming system. In the first scenario, the proposed framework reduced queue length by 60.0%, total delay by 52.9%, and waiting time by 38.0%, while achieving a synchronization accuracy of 76.1%. In the second scenario, the main route queue decreased by 82.8%, average speed increased by 226.2%, CO2 emissions and fuel consumption decreased by 56.3%, and synchronization accuracy reached 95.7%.
These results confirm that the active bidirectional Digital Twin constitutes an effective platform for proactive decision-making, contributing to traffic management that is more efficient, safer, and more sustainable. |
| note de thèses : |
Mémoire de master en informatique |
|  |
A digital twin framework for traffic management
