Research On Curve Trajectory Planning Model:A Driving Simulator Study

With the development of the times and the advancement of technology,more and more intelligent technologies have been applied.Autonomous vehicles have closed to practicality.Trajectory planning,as one of the key technologies for realizing intelligent vehicles,enables vehicles to make decisions on trajectories based on the current road environment.So that the car tracks the trajectory to achieve automatic driving.At present,most trajectory planning methods are constructed based on constraints and objective functions.These methods ignore the driver's driving habits,driving characteristics,and behavioral diversity.Therefore,this article is based on the UC-win Road virtual simulation software,using the driving simulator to collect the trajectory data of the driver in the curve scene.By learning the driving trajectory of real drivers,a "anthropomorphic" trajectory planning model is established.This method is in line with the current general trend of artificial intelligence to achieve autonomous driving.At the same time,the driving trajectory formed by imitating the driver through driving learning and driving experience is considered to be more safe,efficient and comfortable.Intelligent vehicles can provide the same driving experience as a human driver during autonomous driving for passengers.According to the average scene in the curve,based on the analysis of the driver's driving behavior in a curve,an artificial neural network for learning and predicting trajectories is established by selecting appropriate variables.The ANN trains the experimental data,finds the correspondence between the samples and the labels,and then determines the network parameters.Based on the artificial neural network,a curve trajectory planning model that can predict the vehicle path and speed at the next position according to the current road and vehicle information.In addition,the model can continuously update the model parameters according to the trajectory data generated by the vehicle owner to obtain a more personalized trajectory planning model.According to the obstacle avoidance scene in the curve,the obstacle avoidance process is divided into the first lane change process,the intermediate process and the second lane change process.An obstacle avoidance trajectory planning model is established based on the obstacle avoidance process and behavior.The model is divided into three parts,namely a starting point determination module,an intermediate process module and a trajectory decision module.The starting point determination module determines an initial condition according to a threshold value obtained by analyzing an obstacle avoidance starting point of the driver.The intermediate process module determines the relationship formula according to the relationship between the end of the first lane change and the intermediate process distance to obtain the final conditions.The trajectory decision module obtains the lane change trajectory planning model by establishing a neural network based on the initial and final conditions.The lane change trajectory planning model can determine the position of the next moment based on the current vehicle road information,determine the trajectory of the two lane change processes.With these three modules,a complete,safe and continuous obstacle avoidance trajectory can be obtained.