Research On The Decision Model Of Automatic Driving Vehicle Lane Change

In recent years,automatic driving has become one of the important directions of automobile development.Science and technology companies,vehicle manufacturers and scientific research institutions are vigorously studying and developing automatic driving technology,and have achieved certain results.But at present,there are still many problems in the research and development process,among which the problem of automatic driving lane change is one of the most difficult problems to solve.In this paper,a decision-making model based on decision tree is proposed to solve the problem of comparison and key decision-making.The main contents of this paper are as follows:Firstly,get the data of vehicle lane change.The main factors affecting vehicle lane change were determined by consulting data and questionnaire.In order to truly reflect the situation of vehicle lane changing,the real vehicle test is used to collect data.The driving road information,personnel operation information and lane changing information are obtained by determining the scene,personnel and equipment of vehicle driving road,and the data about vehicle influencing factors are collected pertinently.Secondly,the rules of vehicle lane changing based on decision tree are extracted.Through data filtering,noise filtering,lane line recognition and other preprocessing methods,the original data collected by the real vehicle is determined at the beginning and end of lane change,as well as the type of lane change scene.The lane change data corresponding to different vehicle lane change factors in various scenarios are randomly divided into training data and test data.The decision tree algorithm is used to classify the training data,extract the vehicle lane change rules in various scenarios,and verify the feasibility and accuracy of the extracted rules with the test data.Then,the vehicle lane changing model is established.According to the extracted vehicle lane changing rules,use the hierarchical state machine to establish the corresponding top,middle and bottom state machine lane changing rule base.Using Simulink/Stateflow to build the decision-making model of automatic driving lane change.MATLAB/Simulink is used to build input signal model,vehicle control model,output signal model and trigger model,so that these models and lane change decision model constitute a complete vehicle lane change model,so as to simulate and verify the feasibility of automatic driving lane change decision model.Finally,the decision-making model of lane change is simulated and analyzed in virtual scene.In order to reduce the simulation error,VTD software is used to build the simulation traffic scene according to the scene of data collection,and configure the relevant parameters of the experimental vehicle in the simulation scene.Through the joint simulation experiment of the lane change model built by MATLAB/Simulink and the simulation traffic scene built by VTD software,the vehicle parameter change diagram,simulation top view and visual scene diagram in the process of lane change vehicle simulation are obtained,and the comprehensive analysis is carried out.The experimental results verify the feasibility of the decision model based on decision tree.