Socially Smart Connected and Autonomous Vehicle: - Belfort, France - université de technologie de Belfort Montbéliard

Description

Socially smart connected and autonomous vehicle: Deep Reinforcement Learning (DRL) for safe and efficient cooperative intersection:

• Réf

-
ABG-113356

• Sujet de Thèse 13/04/2023
• Financement public/privé université de technologie de Belfort Montbéliard
• Lieu de travail
• Belfort
• Bourgogne-FrancheComté
• France
• Intitulé du sujet
• Socially smart connected and autonomous vehicle: Deep Reinforcement Learning (DRL) for safe and efficient cooperative intersection
• Champs scientifiques
• Informatique

• Mots clés

• Connected and autonomous vehicle, deep reinforcement learning, socially intelligent robots
  Description du sujet:

Intersection management is at the core of studies for improving traffic conditions.

In the half of last century, traffic lights have significantly contributed to the growth of traffic in terms of throughput and velocity through time assignment of the intersection space to conflicting flows.

Initially, designed to improve the safety, they have benefited from several theoretical and technological advances to increase the traffic performance.

However, traffic lights do not fully use the potential of Connected Autonomous Vehicle (CAV).

More precisely, CAVs introduce the following novelties:

• Sequence formation: CAVs can wirelessly negotiate together to determine which CAV crosses the intersection first, which one the second and so on,
• Trajectory planning: Each CAV can autonomously adjust its speed according to the current and future obstacles coming from the other roads.

The thesis work consists in defining in a first step a dynamic driving policy of the vehicle (aggressive, courteous, selfish or grouped...) which is dynamically adapted according to the traffic context and in using in a second step metaheuristics to compute the sequences and train the agents to respect the dynamic sequences.

The result of the thesis can be extended to the cooperation between CAVs and human driven vehicles at intersections.


Prise de fonction:

• 01/10/2023
  Nature du financement:
• Financement public/privé
  Précisions sur le financement:
• Bourse Ministère
  Présentation établissement et labo d'accueil:

- université de technologie de Belfort Montbéliard
Site web:


Intitulé du doctorat:

• Informatique

Pays d'obtention du doctorat:

• France

Ecole doctorale:

• Sciences physiques pour l'ingénieur et microtechniques
• SPIMMaster's degree in computer science,

skilled in oriented object programming languages (mainly Python and C#).

Experiences in DRL are appreciated- 31/05/2023