Research And Implementation Of Platoon Collaboration And Intent Recognition At Intelligent Intersection

Autonomous driving has become a main research topic in the field of intelligent transportation.one of the major challenges is to deal with complicated intersection environments.On one hand,autonomous vehicles cannot accurately predict the driving trajectories of other vehicles when they are crossing the intersections.Therefore,driving safety is threatened.On the other hand,autonomous vehicle may cross intersections with the help of network collaboration,thereby improving traffic efficiency and reducing traffic congestion.In order to solve the above problems,this paper is focusing on trajectory prediction technology of vehicles.Moreover,the problem of autonomous vehicles cooperating through the intersection in the form of platoon is studied.First,this paper proposes a trajectory prediction model derived from GAN(Generative Adversarial Networks)and LSTM(Long Short-Term Memory).A LSTM cell is applied to the generators and the discriminators of GAN,which could encode and decode vehicle trajectory information.In order to ensure the GAN is able to take the interaction among vehicles into consideration when predicting trajectory,this paper proposes a interactive pooling module.This module consists of convolutional layers and pooling layers,which extract the interaction information among vehicles.To evaluate the performance of the proposed model,this paper compares it with other trajectory prediction models like RNN model and LSTM model on the Argoverse dataset.Results show that our model can outperform others with lower mean-square error.In this paper,the formation and splition of cooperative platoon in the intersection scene is studied.Based on the headway between adjacent vehicles and the rapid-exploration random tree(RRT),an algorithm for the formation,splitting of platoon,and path planning is proposed.In addition,this paper proposes a replanning technology,which makes the algorithm adaptive to the dynamic traffic.Finally,a simulation platform is built on the Veins simulation framework,and the performance of the model is evaluated.Simulation results show that the platoon collaboration model in this paper improves the traffic efficiency in the intersections.