The conventional approach to road traffic control encounters several challenges, including inefficiencies in traffic operations, congestion relief, and a lack of overall optimization. Federated Learning presents itself as a promising solution to address these issues facilitating the control of road traffic flow and optimizing collaborative traffic operations among multiple parties, including vehicles and roads. Additionally, Federated Learning offers advantages such as preventing sensitive information leakage and safeguarding data privacy. Through the analysis and extraction of insights from traffic flow data, we develop an optimized control model. To practically demonstrate the feasibility of this approach, we propose to use Federated Learning to build a cooperative vehicular intelligent control system to improve traffic efficiency. The study involved theoretical validation of the system's viability, along with investigations into node selection, network structure design, and algorithm development. The proposed system enables intelligent control by continuously analyzing real-time road conditions. Experimental results substantiate the system's effectiveness in enhancing traffic operation efficiency and improving users' overall travel experience.
Cooperative Vehicular Intelligent Control System based on Federated Learning
31.05.2024
1796773 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Vehicular cooperative perception through action branching and federated reinforcement learning
BASE | 2022
|Vehicular Cooperative Perception Through Action Branching and Federated Reinforcement Learning
ArXiv | 2020
|Federated Learning in Vehicular Networks
IEEE | 2022
|