Sourounded Vehicle Driving Behavior Identification For Decision-making Of An Aumated Vehicle

Recently,with the developing of automatic driving technology,many universities,research institutions,traditional vehicle enterprises,and Internet companies continue to strengthen research and development.However,the automatic driving technology in the real traffic environment still faces many difficulties.And the driving behavior recognition of surrounding vehicles is an important part.In this paper,based on the decision-making requirement of intelligent vehicle,the driving behaviors of environmental vehicles are modeled.The driving behavior recognition of vehicles in most literatures is based on the driver model,in which manipulation data of the drivers and vehicle parameters is used to analyze the driver's intention and predict the vehicles' behavior.Now and in the near future,the vehicle assisted driving or automatic driving vehicle,will run with manned vehicles in real traffic environment,moreover,as connected vehicles are not popular,there maybe no information exchange among vehicles.So in this paper,the automatic driving vehicles with environmental perception sensors,which can detect and track vehicles around(preliminary laboratory results),will recognize these vehicles behavior that will be used in calculation of the vehicle behavior decision making.In this paper,we analysis the driving behavior of the vehicles around the intelligent vehicle in the urban environment and divide the surrounding vehicles into two aspects,the vehicle ahead and the side vehicle ahead.Then,the data acquisition based on the laboratory vehicle platform and Prescan and MATLAB/Simulink joint simulation platform are presented.The vehicle ahead and the side vehicle ahead are studied separately.For the vehicle ahead,a Kalman filter is used to estimate the states of the vehicles.For the side vehicle ahead,we contrast to the selected characteristic factors influence the vehicle driving states.The method of RBF neural network and Support Vector Machine(SVM)is studied.By normalization and continuous adjustment of selecting appropriate parameters,eventually,a target vehicle lateral driving behavior recognition learning model is established.The model can provide the information for the intelligent vehicle decision-making module.Generally speaking,the intention of the vehicle ahead and the side vehicle ahead can be recognized accurately,and the information are sent to the intelligent vehicle decision-making module,avoiding the safety threats caused by the changeable states of environmental vehicles,accordingly improving the understanding and independent ability of the intelligent vehicle in the urban environment.The real-time and effectiveness of the model have been verified on the vehicle.