This study proposes a multi intelligent connected vehicles (ICV) collaborative lane changing model based on Deep Q-Networks (DQN), and verifies its effectiveness in dynamic traffic environment through the test of highway-env simulation environment. The results show that with the increase of training times, the model can significantly improve the overall average speed of the team. Further analysis shows that the model shows different performance under different traffic density and the number of controlled vehicles. The research shows that when the density continues to increase, the reward value, average speed and headway of intelligent connected vehicles all show a downward trend. In addition, when the number of controlled vehicles increases from 3 to 5, although the average reward value and headway of the fleet increase, the average speed decreases. This study provides a new perspective and method for the optimization of vehicle collaborative behavior and the improvement of traffic efficiency in intelligent transportation system.
Intelligent Collaborative Driving Lane Changing Decision of Multi-Intelligent Connected Vehicles
27.07.2024
490069 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Collaborative lane changing decision-making method for intelligent network-connected motorcade
Europäisches Patentamt | 2023
|British Library Conference Proceedings | 2022
|SAE Technical Papers | 2022
|