Research On Prediction Method Of Vehicle Running Track Based On Driver’s Intention Of Lane Change

With the development of intelligent driving technology,the functions of driving aid system in the market are more and more abundant.In order to realize the early warning for the driver’s lane changing operation,it is the key to accurately and realtime identify the driver’s lane changing intention and predict the vehicle lane changing trajectory.Compared with lane keeping,lane changing is a complex driving behavior.As a key component of driving behavior,lane change is closely related to traffic safety and traffic conditions.However,most of the advanced driving assistance systems ignore the driver’s intention of changing lanes.In order to improve the safety of driving and improve the traffic environment,this paper proposes a method combining long short term memory network(LSTM)with evidence theory for driver’s lane change intention recognition,and uses LSTM model to predict driver’s lane change trajectory.The main contents of this paper are as follows:(1)A database of driver’s lane changing intention is established,which consists of3000 groups of lane changing data.The database contains 2500 groups of training data and 500 groups of test data.Through the six degree of freedom driving simulator of Jiangsu University,the driving environment of expressway is simulated,the movement data of driver’s head and eyes are obtained by smart eye Pro eye tracker of Jiangsu University,and the driver’s intention database is established by using the main control unit of driving simulator to synchronously collect the longitudinal acceleration,steering wheel angle and the center distance between vehicle and lane line.(2)A hybrid model of long short term memory network(LSTM)and DS evidence theory was constructed.By comparing and analyzing the loss function and Euclidean distance error in the process of model training under different parameters,the optimal model parameters are selected,and the lstm-ds evidence theory hybrid model with time series and classification characteristics is established.Using tensorflow framework and python language to build the model.(3)Based on the result of driver’s lane changing intention recognition,the vehicle lane changing trajectory is predicted.In this paper,the prediction of vehicle lateral motion track is mainly carried out,that is,the prediction of vehicle lane change.In this paper,a lane change trajectory prediction model based on LSTM is proposed.According to the temporal characteristics of the trajectory,depth learning and evidence theory are applied to the prediction of vehicle trajectory.Considering the traffic scene synthetically,the LSTM deep neural network and the formula of vehicle safety distance are combined to analyze the safety of lane changing.Finally,the feasibility of the method is verified by experiments.The experimental results show that the model can effectively identify the driver’s intention to change lanes.At the same time,the predicted trajectory of the vehicle lane change trajectory prediction model based on LSTM has a high degree of agreement with the real trajectory.