Research On The Vehicle Longitudinal/lateral Driving Safety And Motion Control Based On Inverse Dynamics

As the future development trend of automotive industry,autonomous vehicle is one of the effective ways to improve traffic safety and alleviate traffic congestion.Among the core technologies of autonomous vehicle,automatic driving control technology is a critical issue which ensures the driving safety and stability.The starting point of this thesis is to improve the driving safety under typical working conditions.Aiming at the problem that it is difficult to determine the observation time in the follow-up compensation control,handling inverse dynamics method is used to solve the control input of autonomous vehicle.By applying theoretical analysis,virtual simulation and hardware-in-the-loop test,the modeling and the solving method of inverse dynamics,the safety constraints and control strategies under typical working conditions are thoroughly studied.This research can provide a reference for the development of automatic driving control technology.The main contents lie in the following aspects:1.The modeling of inverse dynamics problem is studied.The force conditions of vehicle during longitudinal and lateral motion are analyzed,and a vehicle dynamic model considering the longitudinal and lateral mechanical characteristics of the tire is established.The control problem of the vehicle is converted into an inverse dynamics problem by adding boundary constraints,process constraints and performance indexes.And then the virtual verification and the hardware-in-the-loop(HIL)tests are implemented to validate the effectiveness of the proposed inverse dynamics problem.The results show that the model demonstrated in this thesis can accurately reflect the characteristics of vehicle longitudinal and lateral motions,and the obtained control input is consistent with the driver’s handling habits.2.The solving method of inverse dynamics problem is studied.This thesis introduces the solving methods of inverse dynamics problem,and summarizes the advantages and disadvantages of several common solving methods through numerical simulations.In view of the fact that the collocation method is prone to produce Runge phenomenon while the calculation efficiency of global pseudospectral method is low when the variation of variables is non-smooth,the local-global adaptive pseudospectral method is proposed to solve the inverse dynamics problem.In addition,aiming at the problem that existing hp-adapticve pseudospectral method ignores whether the constraint requirements are satisfied between the collocation points,an hp-adaptive pseudospectral method that can detect constraints between the collocation points is proposed.3.The driving safety and control input of car-following behavior are studied.The longitudinal car-following safety distance model considering uncertainty is established by applying interval mathematics;and the influence of three typical uncertain factors on safety distance are analyzed.In order to ease the contradiction between the driving safety and traffic efficiency,the minimum safety distance is taken as the car-following distance,and reach the expected car-following state within the minimum time is used as the performance index.By shortening the car-following distance and reaching the car-following state within the minimum time,the contradiction between the driving safety and the traffic efficiency is eased.Meanwhile,considering the riding comfort and the driving state of the leading vehicle,the variable constraint control strategy and the near-optimal feedback control strategy for car-following are established.Finally,the numerical simulations validate the effectiveness of the proposed car-following model and control strategies.4.The driving safety and control input of lane keeping behavior are studied.To generate the target path,a hybrid path planning algorithm is established based on MAKLINK graph theory and improved ant colony algorithm.And then a driving mode recognition model is established based on hidden markov model(HMM)and support vector machine(SVM).According to the performance that passengers are concerned about(i.e.,driving safety,riding comfort and timeliness),three control modes are established,and a multi-mode switching lane keeping control strategy is proposed.Meanwhile,in order to ensure the smooth transition between each mode,a multi-stage inverse dynamic control strategy is established.Then the lane keeping control strategy based on multi-mode switching is verified by the numerical simulations.5.The driving safety and control input of lane changing behavior are studied.Taking the database of next generation simulation(NGSIM)as the sample data,the trajectory data are screened and processed by formulating trajectory screening rules and using exponential smoothing method.Then a lateral safety distance model based on data-driven is established by using deep neural network.The forms of collisions caused by lane changing are analyzed,and the lane changing behavior constraints considering the longitudinal and lateral safety distance are established.In addition,the constraints of overtaking behavior under vehicle networking environment are established,and the influence of disturbance on motorcade is analyzed.Then the simulations under lane changing and overtaking behavior considering the longitudinal and lateral safety distance are carried out.The results show that the proposed control method can meet the requirement for safety distance in the longitudinal and lateral directions during the process of lane changing or overtaking.