Research On Driver-Vehicle-Road Cooperative Adaptive Hierarchical Collision Warning Algorithm In The Internet Of Vehicles Environment

A new round of technological revolution and industrial transformation is in a new period of vigorous development.As a key carrier of innovation and integration of automobile and transportation,energy,communication and other fields,intelligent connected vehicles are gradually becoming the strategic commanding point of the development of the global automobile industry.At the same time,with the cooperation and development of artificial intelligence,visual computing,radar perception and global positioning system,intelligent connected vehicles are marching towards driving automation and unmanned.However,the autonomous driving technology in the Internet of vehicles environment is still in the experimental stage,and vehicle active safety in complex traffic driving environment is always a hot and difficult research topic.Therefore,in the context of the development of the intelligent connected vehicles active safety technology,a driver-vehicle-road cooperative adaptive hierarchical collision warning algorithm in the Internet of vehicles environment is proposed by considering the vehicle state estimation under complex driving conditions,the tire-road adhesion coefficient estimation under different road conditions and the driver’s driving intention recognition under different relative driving conditions.The algorithm is helpful to improve the safety and comfort of intelligent vehicles with auxiliary active safety system.The main contents are as follows:(1)In order to accurately obtain the longitudinal velocity and position of the vehicle in the Internet of vehicles environment,a vehicle motion state estimation algorithm based on vehicle nonlinear kinematics model in complex traffic environment is proposed.Considering the real-time interaction of vehicle-vehicle information can be realized in the Internet of vehicles environment,and the relative motion states of the front vehicle and the rear vehicle can be characterized by vehicle state information such as the relative distance and relative velocity of two vehicles,a vehicle nonlinear kinematic model including the midpoint position of front axle and rear axle is established.By designing adaptive algorithms of process noise and observation noise based on Extended Kalman Filtering(EKF),the interference of information acquisition errors of vehicle sensors in complex traffic environment can be effectively compensated,and the practical problem that it is difficult to obtain the statistical characteristics of prior noise accurately can be solved.The difference between the established vehicle kinematics model and the actual physical process may affect the filtering ability of the EKF in processing vehicle data information,and even lead to the divergence of the filtering.Therefore,an improved EKF algorithm combining the limited memory filtering and random weighting theory is proposed.By selecting observations of a certain length memory interval and introducing the weighted coefficient σ_i of the observation noise matrix R,the algorithm can detect and adjust the observation noise in real time,enhance the correction effect of new measurement information in the filtering estimation at the next moment effectively,reduce the dependence on model parameters in the filtering process and suppress the divergence of filtering.At the same time,a MATLAB/Simulink and Carsim co-simulation platform is built to carry out the simulation calculation in complex traffic driving environment to verify the validity and accuracy of the algorithm.(2)It is necessary to estimate the maximum deceleration that the vehicle can achieve before the Automatic Emergency Braking(AEB)system is implemented,so as to accurately judge the safe distance between two vehicles.A multi-condition tire-road adhesion coefficient estimation algorithm based on vehicle nonlinear dynamics model is proposed.Firstly,a vehicle nonlinear dynamics model with longitudinal,lateral and yaw directions is established.Secondly,the longitudinal force and lateral force of the tire are normalized by analyzing the Dugoff tire model.Then,considering the complexity of the algorithm,an improved Unscented Kalman Filtering(UKF)algorithm based on the fusion of limited memory filtering and random weighting theory is used to estimate the tire-road adhesion coefficient.Finally,a co-simulation platform of MATLAB/Simulink and Carsim is built to validate the algorithm under different conditions such as high adhesion road,low adhesion road,split-μ road and variable high/low adhesion road.(3)Considering the influence of driving intention on vehicle safety,the longitudinal driving process of vehicle is described as an uncertain hidden Markov random process in accordance with time series.The model divides the driving intention into constant speed,acceleration,deceleration and emergency braking.At the same time,according to the relative distance and relative velocity between two vehicles,nine kinds of vehicle relative driving states are listed,and the mapping relationship between observable vehicle relative driving state and hidden driving intention is established by using Hidden Markov Model(HMM).The driver-in-the-loop co-simulation experiment platform is built for data acquisition.T-test algorithm is used to preprocess the data,and the processed sample data are classified for training and testing of driving intention recognition model.In the meanwhile,based on the vehicle safety distance theory,the vehicle motion state,the tire-road adhesion coefficient and driving intention are introduced into the construction of collision warning algorithm,and the driver-vehicle-road cooperative adaptive hierarchical collision warning algorithm is designed in the Internet of vehicles environment.Finally,the effectiveness and accuracy of the algorithm are verified by comparing with the traditional Mazda model under four different driving intentions.