Research On Control Strategy Of Lateral Assistance Driving And Man-machine Mutual Driving

As the number of vehicles has increased dramatically,it brings a series of new problems,such as frequent occurrence of traffic accidents.In order to reduce the traffic accidents caused by the human error,ADAS has become a research focus in recent years;lateral assistance driving system is an important part of ADAS.At present most lateral assistance driving system is steering angle closed-loop control based on linear vehicle-road model regardless driver in the loop,if the assistance sysem unnecessary intervention,it will seriously affect the driver trust in the system.Therefore,steering angle closed-loop control based on time-varying vehicle-road model,guaranteed cost coordinated control based on steering system and braking system,driver-vehicle-road closed loop control based on steering torque input and properly identifying driver's intention man-machine mutual driving cooperative control of the lateral assistance driving remain to be further research.In this paper,the control strategy of lateral assistance driving system in the domestic and overseas are analyzed and summarized,and based on CarSim/Simulink and CarSim/LabVIEW RT,different control strategies of the lateral assistance driving on the theoretical analysis,numerical simulation and hardware-in-loop simulation are studied.The main work and achievements are as follows:1.Based on path error vehicle-road model a switched steering control strategy is proposed by a local model with time-varying vehicle speed as sub models of lateral assistance driving.The switched steering controller parameters are acquired by the state feedback ?-suboptimal H_? norm and linear matrix inequalities.The lane keeping performance of the designed switched steering controller is compared with a steering controller designed based on L2 WVM,the robustness of the lane keep switched steering controller is verified by changing the quality and yawing moment of inertia of the vehicle.A tracking path vehicle-road model of the lateral assistance driving is established,the model predictive steering controllers are both designed by the path error vehicle-road model and the tracking path vehicle-road model,based on CarSim/Simulink the lane keeping performance of the designed model predictive steering controllers are compared.2.Based on lateral offset in the single-point preview and two degrees of freedom vehicle dynamic model lane keeping desired side slip angle and desired yaw rate are acquired.The steering angle and corrected yawing moment to improve vehicle path tracking ability and lateral stability are generated by an upper guaranteed cost controller,based on vehicle stability index a monitor is designed to coordinate the steering and braking system intervention.Based on vehicle dynamic model a side slip angle estimation algorithm is amended by the side slip angular velocity,and it is concluded that the algorithm has higher estimation accuracy and robustness,the lane keeping performance of the guaranteed cost coordinated control is verified.A T-S fuzzy model with steering torque input of the lateral assistance driving is established,a lane keeping PDC/H_? disturbance rejection steering controller of the assistance system is designed.The simulation results show that based on driver-vehicle-road closed loop model the steering controller improves the lane keeping performance and has less conflict between the driver and the steering controller.3.The driver intention recognition algorithms are proposed based on the on-center handling characteristics threshold and the D-S evidence theory,which basic trust function assignment is acquired by BP neural network.The simulation results show that the method based on D-S evidence theory can more timely recognize driver intention based on CarSim/Simulink.A vehicle-road model based on steering wheel angular velocity input is established,and an upper sliding steering controller is designed,and a desired yaw rate observer is designed by vehicle kinematics model,the simulation results show that based on the proposed desired yaw rate observer response of the lane departure avoidance system is fast,and the stability is good.Based on an equivalent dynamic model of vehicle steering system a lower torque sliding mode controller is designed.Because the relationship between the lane departure degree with a lateral offset in the single-point preview and the driver torque cannot be accurately described,a weight coefficient fuzzy observer for man-machine mutual driving cooperative control is designed.The results show that the proposed man-machine mutual driving cooperative control strategy is able to return the vehicle to the normal lane in a timely fashion based on CarSim/Simulink.4.In order to realize smooth switching from man-machine mutual driving to driver control,switching criteria from man-machine mutual driving to driver control is proposed.In order to ensure that the system is able to correctly return the vehicle to the normal lane under complicated driving conditions,while considering the uncertainty of the time varying vehicle speed and tire cornering stiffness,a gain scheduling brake controller is designed based on the energy-to-peak performance indicator.The hardware-in-the-loop test platforms for the lateral assistance driving and man-machine mutual driving are set up based on EPS and ESP systems,and the effectiveness of lane keeping guaranteed cost coordinated control strategy and lane departure avoidance steering control strategy using driver's intention recognition are both verified based on CarSim/LabVIEW RT,the ability to correctly return the vehicle to the normal lane by man-machine mutual driving with EPS system is compared with EPS and ESP systems.The results show that the proposed man-machine mutual driving cooperative control strategy is able to return the vehicle to the normal lane in a timely fashion,could realize smooth switching from man-machine mutual driving to driver control,and avoid inadvertent lane departures.Finally,the research contents are summarized and points out the deficiency of this topic research,a vehicle lateral assistance driving tested platform is intend to set up by a pure electric vehicle,which equips with a lane recognition system based on machine vision,and through a real car test to evaluate performance of the proposed lateral assistance driving and man-machine driving cooperative control strategy.Due to the driver intention recognition depending on the collected information from various sensors,and in order to solve possible conflict for fusion all the evidence,weight of different evidence will be used to improve driver intention recognition algorithm.