Recent developments in Cooperative Intelligent Transport Systems (C‐ITS) have focused on autonomous driving as an emerging technology capable of reducing accidents by improving road safety and respecting driving rules. Through the use of vehicle‐to‐everything (V2X) communication technologies, C‐ITS enable connected autonomous vehicles to share information with each other, as well as with infrastructure and other connected road users. This chapter presents a method of collective perception based on multi‐agent deep reinforcement learning. This method enables each connected autonomous vehicle, treated as an agent, to learn an independent interactive strategy that will be used to select and exchange detected objects in the network. The results of the simulations show that the proposed method, compared to the current works of the state of the art, improves network reliability without sacrificing the level of sensitivity within short ranges of under 100 meters.
Resilience of Collective Perception in C‐ITS – Deep Multi‐Agent Reinforcement Learning
2024-10-25
21 pages
Article/Chapter (Book)
Electronic Resource
English
Cooperative perception in vehicular networks using multi-agent reinforcement learning
BASE | 2021
|Multi-Agent Deep Reinforcement Learning in Vehicular OCC
IEEE | 2022
|