Self-Organizing Traffic Based on Dynamic Platoon Configuration

Abstract

Nowadays, urban environments suffer from recurrent traffic jams and their associated side effects. In order to cope with this issue, several solutions were proposed, mostly relying on road-operated traffic management strategies. Recent studies show the potential of connected and automated cars to improve traffic by adapting individual behaviors. For instance, thanks to autonomous decision-making capabilities, platoon systems enable a group of vehicles to travel attached to each other by a virtual link. Multi-Agent Systems (MAS) empowered by Cyber Physical Systems (CPS) can be used to control a fleet of cars in a decentralized way. This paper applies this concept for dynamic platoon vehicles system through a multi-agent model based on Newtonian physics to achieve a self-organization at the. The behavior can be improved by offering more computing power or a better knowledge of the environment thanks to a micro-service division. In this proposal, the vehicle is considered to be a physical agent with personal behavior divided into three layers : perception based on sensor, decision based on physics simulation and application helped by embedded robotics.