Cité Descartes, Champs sur Marne
77447, Cedex, France.


Ifsttar Batiment Bienvenüe
Ecole des Ponts ParisTech (ENPC)


N. Farhi, N. Bhouri et J-P. Lebacque (Grettia).
Z. Christoforou et F. Leurent (Lvmt).


+331 81 66 87 04

Session prochaine

Date: 17 Novembre 2016
Lieu: Ifsttar - Amphi A017 - 14 - 20 Boulevard Newton - Marne-la-Vallée, France. Accès .


14h00 Antoine Tordeux, Forschungszentrum Julich GmbH and Bergische Universitat Wuppertal, Germany.
Titre. Motion planning and control techniques for driver assistance systems and autonomous vehicles.
Résumé. Road driving assistance and automation systems are nowadays extensively developed in research centres and universities, and start to be commonly proposed by manufacturers. Partial and full automated driving are subject to speculation on a new quality of road transportation in terms of safety, comfort, mobility, performance and environment. We propose to review the literature of connected and autonomous vehicles, automated motion planning techniques and associated control and safety processes.
Mots clefs. Road transportation; Driving assistance and automation systems; Connected and autonomous vehicles; Motion planning techniques; Traffic control and safety.
14h45 Walid Behiri et Sana Berraf, Esiee.
Titre. Transport ferroviaire urbain de marchandises dans la perspective de mixité fret et voyageurs : modélisation et optimisation.
Résumé. This work studies an alternative way of transporting freight by using urban rail infrastructure. First, we identify and classify the different possibilities of mixing freight and passengers using rail network. As a second contribution, several optimization problems are identified dealing with urban freight. In this study, the addressed problem deals with a commuter line on which each station can be used as a loading/unloading platform for goods such as demand (goods boxes) is known in advance. As a third contribution, a discrete event simulation model is developed to evaluate the proposed transport solution, where the operational transport of goods is performed using several heuristics. Finally, a Mixed Integer Linear Program (MILP) model is proposed to optimize the process of goods transporting, with an objective of minimizing the total waiting time of daily deliveries such that each box is transported from its departure station to its arrival station. Numerical results show that a MILP model is able to solve instances whose size is close to realistic case in very short amount of time.
15h30 Pause Café.
15h45 Farida Manseur, Grettia, Ifsttar.
Titre. Robust adaptive strategies for the guidance of users in road networks.
Résumé. We present an algorithm for optimal guidance of users in road networks. It is a "stochastic-on-time-arrival (SOTA)"-like algorithm which calculates optimal guidance strategies with reliable paths, for road network origin-destination pairs. Our contribution consists here in extending an existing SOTA algorithm, in order to include robustness of the guidance strategy, towards path failures. We take into account here the existence and the performance of alternative detours of the selected paths, in the calculus of the guidance strategy. We take into account the fact that one or many links of the selected optimal path may fail during the travel. We then consider that users may be sensitive to path changing. That is to say that they may prefer paths with efficient alternative detours, with respect to paths without, or with less efficient detours, even with a loss in the average travel time, and/or in its reliability. We show the effectiveness of our algorithm on a grid network, where the vector of the travel times through the links of the network is assumed to have a multivariate gamma probability distribution, taking into account the correlations of the travel times.
16h30 Kwami Sossoe, Grettia, Ifsttar.
Titre. Reactive Dynamic Assignment in Discrete-continuous Large-scale Transport networks.
Résumé. In the field of traffic flow modeling of large-scale surface networks, a bi-dimensional traffic flow modeling framework emerged. It is a family model developed at a greater scale. The theory of traffic flow at this scale of modeling is a high-level of aggregation of road sections and road intersections, in zones of two dimensions. The corresponding network domain of such surface network is approached as a continuous network that comprises infinite secondary roads. A specific dynamic network loading is proposed to compute vehicles flow on such continuous network, while a reactive assignment scheme is designed for dynamic traffic assignment. A Dijkstra-based algorithm is used to find time dependent shortest paths over large networks in this context, where instantaneous travel times are considered instead. Since urban network may comprise major roads like highways, a coupling of GSOM macroscopic flow model (applied to discrete network) with the bi-dimensional flow model (applied to the continuous network) is discussed. The new proposed transport model has the capability of reducing significantly computational complexity. Moreover, the model is suitable for networks with few traffic data sources.