Skip to Main content Skip to Navigation
New interface

Implementation of a cooperative communication within a fleet of connected and autonomous vehicles

Guilhem Marcillaud 1 
1 IRIT-SMAC - Systèmes Multi-Agents Coopératifs
IRIT - Institut de recherche en informatique de Toulouse
Abstract : This thesis addresses the problem of communication in the context of a fleet of autonomous and connected vehicles. It is included in the context of intelligent transport systems and the smart city. Numerous applications have been developed in recent years in these areas, each with the objective of improving the quality of life of users. The automation of driving, which has been a central concern of the automotive industry since the 2000s, is expected to reduce the number of accidents, improve the comfort of users and also reduce the ecological footprint of road traffic in general. It also paves the way for the implementation of effective cooperative strategies between vehicles. As cooperation relies on the exchange of relevant information, it is necessary that vehicles are able to know what information to transmit and how. This thesis focuses on inter-vehicle communication and approaches the problem as a complex system in which many vehicles interact with each other, each has its own local objective. Each vehicle perceives information in its local environment and knows that some of this information may be useful to neighbouring vehicles. As a cooperative entity, it will share with its neighbours the information it assumes to be useful. In this work, an inter-vehicle communication is considered useful if it verifies the following two properties: 1) the information exchanged is understood by the receiving vehicle and 2) it brings new knowledge to it. In the context of a fleet of Autonomous and Connected Vehicles (CAVs), these two properties may not always be guaranteed, especially if the vehicles involved do not share the same referential frame (e.g. different units of measurement for the same information) or if the volume of communication exceeds the vehicle's capacity. The contribution of this thesis is twofold: it proposes a first module allowing a vehicle to adapt information to its own reference system, and a second module allowing to optimise information exchanges within a fleet of CAVs. It approaches the problem of a common referential frame between vehicles as a data estimation problem, and that of the optimisation of information exchanges between vehicles as a distributed optimisation problem under constraints. The originality of this work lies in the use of adaptive multi-agent systems (AMAS) to solve them. The AMAS approach is an organisational approach to building complex systems that adapt, continuously and locally, to the dynamics of their environment. It focuses on the interactions between the system and its environment on the one hand and between the parts (agents) of the system on the other. These interactions are based on local and cooperative processing of information by the parts of the system, which only have a partial view of their environment. This principle of locality guarantees the emergent nature of the system's operation. The evaluation of the two modules was carried out using various datasets highlighting disturbances that could affect the system (modification of the environment and intermittence of vehicles in the fleet). The results show that both modules are effective for large-scale problems in a dynamic environment. The use of a local approach to solve the problem avoids an exponential increase in complexity. In the context of optimising the information exchanged, and in order to propose a solution that preserves the confidentiality of the data, the local solution of the problem is not based on the exchange of the CAVs' personal information.[...]
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Monday, July 18, 2022 - 3:28:40 PM
Last modification on : Friday, August 26, 2022 - 3:51:12 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03726474, version 1


Guilhem Marcillaud. Implementation of a cooperative communication within a fleet of connected and autonomous vehicles. Artificial Intelligence [cs.AI]. Université Paul Sabatier - Toulouse III, 2022. English. ⟨NNT : 2022TOU30076⟩. ⟨tel-03726474⟩



Record views


Files downloads