Research On Curve Change Control Technology Of Driverless Vehicle Based On Model Prediction

With the updates and iterations of vehicle engineering technology and information technology,vehicles are entering the era of unmanned driving,and the research on lane-changing trajectory planning and tracking control of unmanned vehicles can effectively avoid traffic accidents caused by lane-changing.In this paper,for the high-speed long-curve driving environment,corresponding lane-changing decision-making strategies are developed according to different driving conditions,and research on trajectory planning and tracking control is carried out,including:Firstly,based on the long short-term memory(LSTM)neural network,this paper constructs a surrounding vehicle trajectory prediction model.By tuning parameters such as the number of training data samples and the number of hidden layer neurons in the neural network model,a three-layer LSTM neural network model is finally constructed.The vehicle trajectory is divided into training set,validation set,and prediction set,and the prediction results show that the model can effectively predict the trajectory of surrounding vehicles,with an accuracy of 94.85%.In addition,the MSE results of the horizontal and vertical coordinates are 0.1709 and 0.02553,respectively,and the MAE results are 0.043 and 0.087.The LSTM neural network model constructed has high accuracy and reliability in predicting the trajectory of surrounding vehicles for autonomous driving.Secondly,a decision-making model for automatic lane changing of unmanned vehicles is constructed based on fuzzy logic control,and the lane changing decision of unmanned vehicles is divided into two forms: free lane changing and forced lane changing.The speed difference coefficient and expected vehicle distance are used as the inputs of the free lane changing decision-making model,and the speed difference coefficient,vehicle distance,and the difference between the minimum safe distance for lane changing are used as the inputs of the forced lane changing decision-making model.Based on fuzzy control,a lane-changing decision-making model is constructed with lane-changing willingness as the output.Thirdly,in the automatic lane-changing scenario of high-speed curved roads,the vehicle’s position in Cartesian coordinates is transformed into the position in Frenet coordinates by coordinate transformation.Trajectory planning is performed in Frenet coordinates,and the lanechanging trajectory is decoupled into horizontal and vertical trajectory planning based on the fifth-order polynomial lane-changing trajectory model.The planned horizontal and vertical vehicle position coordinates are spliced into the trajectory of automatic lane changing of unmanned vehicles through smoothing algorithms.At the same time,based on a simplified vehicle dynamic model,a linear error model and a prediction model of the unmanned vehicle control system are established.Based on quadratic programming optimization objective function and adding constraints such as front wheel steering angle and front wheel steering angle increment,a model-predictive trajectory tracking controller is constructed.The controller shows strong performance and can ensure the unmanned vehicle accurately tracks the desired trajectory while considering practical control constraints such as front wheel steering angle and front wheel steering angle increment.Fourthly,through joint simulation of MATLAB and CarSim,this study verified the constructed automatic lane-changing decision-making model under three driving conditions: the front vehicle is driving at a constant speed,the front vehicle is braking urgently,and there is a vehicle ahead in the target lane.The results show that the model performs well in terms of safety and effectiveness.The fifth-order polynomial trajectory planning method based on Frenet coordinates successfully plans the lane-changing trajectory.At the same time,the modelpredictive trajectory tracking controller has strong robustness,can accurately control the unmanned vehicle to track the desired trajectory,and the vehicle’s lateral acceleration and yaw rate are within ±0.4g and ±10deg/s,ensuring the safety and comfort of passengers.