Description du projet
Un système de modélisation adaptative pour le transport durable
Faute d’une approche de modélisation globale, la mise en place de systèmes de transports efficaces et durables reste un défi majeur. Le projet ADAPT-OR, financé par le CER, aborde cette question en proposant un cadre de modélisation holistique et adaptatif. Pour cela, il tient compte de l’interaction entre l’offre et la demande à différents niveaux, ce qui favorise une prise de décision flexible. Le projet vise à améliorer l’efficacité et la durabilité en développant des méthodes d’optimisation et d’apprentissage adaptatives qui permettront d’ajuster dynamiquement le dimensionnement de la flotte, la capacité des installations et les décisions de transport en fonction des performances opérationnelles et des préférences des utilisateurs. ADAPT-OR s’efforcera de doter les chercheurs dans le domaine des transports d’algorithmes d’auto-apprentissage, ce qui permettra aux fournisseurs de services d’améliorer leurs activités, de proposer de meilleures prestations et de concevoir des solutions de transport plus durables.
Objectif
Transport systems are complex with various entities across different decision-making levels. There is currently no comprehensive way to model these entities and their interactions, which prevents utilizing the full potential of the system. For efficient and sustainable transportation, we need to model the perspective of both the supply (e.g. services, infrastructure) and the demand side (e.g. preferences of users). Only then a holistic modelling framework can be developed where the decisions at different levels learn from one another and are adapted continuously in a robust way while accommodating the different preferences.
I propose a holistic adaptive modelling framework that considers the interaction between different levels, both on the supply and demand sides, in order to adapt the decisions towards increased efficiency and sustainability. This necessitates a paradigm change in modelling as it is challenging to maintain robustness across different time-scales at the network level. Even though integrated models for multiple decision-making levels (strategic, tactical, operational) are a trend, they only allow a reactive ex-post assessment but are not dynamically coupled (not self-learning). I plan to achieve this by developing model-based adaptive optimization and learning methods with my expertise on optimization and behavioural modelling. For example, based on the performance of the routing decisions at the operational level in terms of delays, the decisions on fleet sizing and/or capacity of facilities will be adapted. Similarly, based on a continuous learning of the preferences of users, transport decisions will be adapted.
ADAPT-OR will lead to new models and algorithms for transportation researchers (and beyond) with self-learning capabilities. This capability will enable service providers to adapt and maintain their business, users to receive better services and society to reach sustainable transport solutions addressing one of EUs grand challenges.
Programme(s)
- HORIZON.1.1 - European Research Council (ERC) Main Programme
Thème(s)
Appel à propositions
(s’ouvre dans une nouvelle fenêtre) ERC-2023-STG
Voir d’autres projets de cet appelRégime de financement
HORIZON-ERC - HORIZON ERC GrantsInstitution d’accueil
2628 CN Delft
Pays-Bas