Research On Driving Scheme Modeling And Parameter Identification In Turning Condition

Autonomous driving is an important development direction of the automobile industry.The development of autonomous driving assistance system has been relatively perfect,and the world’s major research institutions are making full efforts to develop autonomous vehicles.In the process of full autonomous driving or process autonomous driving level L3 automation,the driving scheme adopted by different drivers under various driving conditions(the change law of vehicle position and speed when human drivers are driving the car)is one of the key research and development contents.One of the key problems is to describe the driving schemes of different drivers and to reflect the driving habits of different drivers.In this paper,a parametric modeling of a driving scheme and a method of model parameter identification in the experiment are proposed for turning conditions.The model is used to describe the driving scheme,and the driving habits of different drivers are reflected by the model parameters.The fitting method is used to obtain the model parameters.Most of the current studies on driving scheme models are used for following control,that is,a fixed trajectory and corresponding speed change are generated after the input of traffic scenarios and driving environment,and insufficient consideration has been given to some parametric models associated with driving habits.The driving scheme model under turning conditions established in this paper is divided into two parts: shift and speed.The shift part is used to describe the position change law of the driving scheme,and the speed part describes the speed change law of the driving scheme.The combination of the two forms a complete driving scheme.The model parameters were obtained by fitting the driving scheme model with the original driving data.The model parameters represented the driving habits to some extent,that is,the turning speed and turning radius used by different drivers or driver groups in turning conditions,thus achieving the extraction of driving habits in turning conditions.This paper focuses on three aspects:First,the establishment of parametric driving scheme model under turning conditions.In this paper,a three-stage model is proposed.The two straight sections at the entrance and exit of the turn are the preparation sections for the turn,which are given by the traffic task.We extract the turning radius and turning speed of the arc segment to represent the style characteristics.Different driving habits will have different turning radius and speed through the arc segment.The main section under the turning condition is the arc section,and the other two straight sections are the buffer section,which is used to change the speed to adapt to the turning requirements and achieve the desired turning speed.Second,the fitting of driving scheme model parameters.In this paper,not only the above two characteristic parameters should be fitted,but also the position of the inlet,the direction of the inlet,the position of the outlet,the direction of the outlet,the velocity of the inlet and the velocity of the outlet should be fitted together.However,the variation degree of the fitting parameters,such as the velocity of the inlet and the velocity of the outlet,is not obvious,so the nonlinear least square method is adopted for fitting.Third,preliminary exploration of driving mode model.After the characteristic parameters were obtained by fitting the driving data of a single driver,the arc radius and arc speed were taken as two-dimensional data for cluster analysis,and the driving mode model of a single driver was established based on the clustering results,which could reflect the driving habit ratio of the driver under different turning conditions.The same method is used to establish a driver group driving mode model for the laboratory driver group,which reflects the proportion of driving habits of the laboratory driver group.The experimental results show that the model and fitting method established in this paper converge to the fitting variance,and the model and fitting method are feasible.The model and parameter identification method established in this paper can be used to obtain the driving habits of drivers and driver groups in the laboratory under different turning conditions.When applied to the automatic driving system,the parameters identified in this paper can be used to conduct personalized configuration for the driving mode of laboratory drivers or laboratory driver groups.