Research On Lateral Auxiliary Control Strategy Based On Prediction Of Driving Behaviors

Traditional passive safety technology through seat belts and airbags can reduce injury and loss to a large degree in the accidents,but has been difficult to meet people’s growing demand on the traffic safety.And automotive active safety technology has raised widely and lasting concern because it can take positive measures to avoid danger and protect people’s life and property safety at its source.Further more,as a typical active safety technology,the horizontal auxiliary system focus on keeping vehicle running in the driveway so as to ensure the vehicle safe.Supported by the National Natural Science Fund Project "Research on the steering control mechanism and evaluation method of information system based on the characteristics of drivers by wire"(Number: 51575223)and commiting to the contradiction between fixed model parameters of auxiliary system and drivers’ diversity demands,this paper proposes a data-based lane departure determination method and introduces MPC(Model Predictive Control)algorithm to active steering control.Finally,the effectiveness of the warning and active control algorithm are validated by the CarSim with Matlab/Simulink simulation and hardware in the loop test bench experiments.Main research content is as following:(1)Build the driving behavior prediction model.Based on the theory of Hidden Markov Model,the identification model is built to predict the most likely driving behavior in the future.At the same time,taking the identification results of current operating state into consideration and concering the shortcoming of the threshold-based algorithms,this paper proposes the data-based method to determine lane departure.(2)Study the effectiveness of proposed warning algorithm.In order to alleviate humanmachine conflict,the current operating state of driver must be identified on the basis of driving behavior prediction.The results from above codetermine the warning trigger point.In addition,for validating the effectiveness and advancement of the proposed warning algorithm,a comparison model based on Time to Lane Crossing is built under straight and curve working conditions.Then simulation experiments are carried out and the results show that the data-based method can early detect the driver’s unconscious lane deviation and give driver more reaction time.(3)Research on active intervention control algorithm.This paper puts forward the strategy of active steering control on the basis of the above identification results.Besides,this article adopts the method of virtual lane line to determine the different intervention time points for different types of drivers.Further,MPC algorithm is introduced to active steering control and simulation experiments based on Car Sim and Matlab/Simulink are done,whose results show that,in this paper,based-on MPC algorithm for lane keeping active control can effectively prevent the vehicle out of the driveway(4)Carry out hardware in loop experiments.Considering the gap between pure software simulation and real vehicle experiment,this paper builds horizontal auxiliary hardware in loop test bench based on EPS to further demonstrate the effectiveness of lateral auxiliary control algorithms,and gives a brief introduction about the platform design,system composition,part of the sensor test and calibration.Finally,the hardware in the loop experiments of auxiliary control algorithms are carried out and the bench test shows that the control algorithms developed in this paper can correct the deviation trajectory very well and maintain the vehicle in the original laneso as to effectively avoid the lane deviation accident.