Vous êtes ici : GIPSA-lab > Formation > Thèses soutenues
TUMASH Liudmila

Traffic control in large-scale urban networks

 

Directeur de thèse :     Carlos CANUDAS-DE-WIT

Co-directeur de thèse :     Maria Laura DELLE MONACHE

École doctorale : Electronique, electrotechnique, automatique, traitement du signal (EEATS)

Spécialité : Automatique Traitement du signal et des images

Structure de rattachement : CNRS

Établissement d'origine : Technische Universität Berlin (TU Berlin)

Financement(s) : Erc ; Sans financement

 

Date d'entrée en thèse : 01/09/2018

Date de soutenance : 07/09/2021

 

Composition du jury :
PRIEUR Christophe, Directeur de recherche, GIPSA-Lab, Grenoble INP, Président du jury.
BEKIARIS-LIBERIS Nikolaos, Professeur assistant, Technical University of Crete, Rapporteur
CLAUDEL Christian, Professeur associé, University of Texas at Austin, Rapporteur.
LECLERCQ Ludovic, Directeur de recherche, Université Gustave Eiffel, Examinateur.
PICCOLI Benedetto, Professeur des Universités, Rutgers University-Camden, Examinateur.

 

Résumé : This research is done in the context of European Research Council’s Advanced Grant project Scale-FreeBack. The aim of Scale-FreeBack project is to develop a holistic scale-free control approach to complex systems, and to set new foundations for a theory dealing with complex physical networks with arbitrary dimension. One particular case is intelligent transportation systems that are capable to prevent the occurrence of congestions in rush hours. The contributions of the present PhD work are mainly related to boundary control design and modeling of traffic on large-scale urban networks. We consider traffic from the macroscopic viewpoint describing it in terms of aggregated variables such as flow and density of vehicles. The corresponding dynamic equation corresponds to a first-order hyperbolic partial differential equation. Within this PhD thesis, we propose control design techniques that rely on the intrinsic properties of the model. First of all, we solve one-dimensional (1D) boundary control problems, i.e., for traffic evolving on single roads. Thereby, the traffic state is driven to a space- and time-dependent desired trajectory that admits traffic regimes switching. Such a control design is far from being trivial due to nonlinearities of the state equation.
Then, the problem is extended from one single road to urban networks of arbitrary size. The large-scale traffic dynamics are described by a two-dimensional (2D) conservation law model. The model parameters are defined everywhere in the continuum plane from its values on physical roads that are further interpolated. The traffic flow direction is determined by network geometry and infrastructure parameters. This 2D model is applicable to any urban area with a preferred direction of motion. For this case, we elaborate a unique method that considerably simplifies control design for urban traffic systems. In particular, we present a curvilinear coordinate transformation that translates a 2D continuous traffic model into a parametrized set of 1D systems. This enables an explicit elaboration of strategies for various control tasks to solve on large-scale networks: we calculate steady-states, design boundary control, as well as apply variable speed limit control.
Finally, a new multi-directional 2D continuous traffic model is presented. This model is formally derived from the demand-supply concept at one intersection. It is called the NSWE model, since it consists of four PDEs that describe the evolution of vehicle density with respect to cardinal directions: North, South, West and East. The traffic flow direction is determined by turning ratios at intersections. We then design a boundary control that drives a multi-directional congested traffic to a desired equilibrium. The effectiveness of our contributions were tested using simulated and real data. In the first case, the results are verified by using the well-known commercial traffic Aimsun, which produces microsimulations of vehicles’ trajectories in a modeled network. In the second case, the real data are obtained from sensors located in the downtown area of the city of Grenoble and collected using the Grenoble Traffic Lab (GTL).


GIPSA-lab, 11 rue des Mathématiques, Grenoble Campus BP46, F-38402 SAINT MARTIN D'HERES CEDEX - 33 (0)4 76 82 71 31