Research On Intelligent Vehicle Motion Control For Variable Adhesion Coefficient Roads

Motion control is an important part of driverless technology.The road conditions may change at any time during the actual driving process,and the tire-ground friction characteristics will change accordingly.Therefore,it is necessary to develop a motion control algorithm that can adapt to different road conditions.There are significant differences in the dynamic characteristics of different parts of the vehicle.It is difficult to achieve control convergence using an integrated control strategy,because it cannot meet the control requirements of each subsystem at the same time.Aiming at the road conditions with varying adhesion coefficients,this research is devoted to adopting a reasonable layered control structure to realize the safe and efficient horizontal and vertical trajectory tracking control of intelligent vehicles.(1)The reference model is the basis for control decisions.The research needs a model that can accurately reproduce the dynamic characteristics of the vehicle under different road conditions.By weighing the fitting accuracy and calculation efficiency,this paper finally chooses 4 degrees of freedom which are directly related to motion control to build the body model.Combined with the MF-Swift tire model which is extended adhesion coefficient interface,it forms a complete vehicle dynamics model for variable adhesion coefficient road conditions.Road condition recognition is the prerequisite for adaptive control of road conditions.In order to maximize the reuse rate of vehicle model and control system sensors,an untracked Kalman filter algorithm based on vehicle dynamic response is established to estimate the road adhesion coefficient.In order to prove the effectiveness of the estimation algorithm,this paper carried out experimental verification under the scenario where the road adhesion coefficient changes in various forms.(2)The form of control strategy adopted is the core factor that will affect the accuracy of motion control.The analytical solution from the target trajectory to the vehicle's desired steering parameters does not necessarily exist.Aiming at the difference in model characteristics between the body system and the tire system,a trajectory tracking layered control strategy is established.Based on the body system model has the characteristics of clear structure,the upper layer used sliding mode control to solve the front wheel lateral and longitudinal forces required for trajectory tracking control.The tire system has strong non-linear characteristics and different dynamic response characteristics under different road conditions.The research collected the input and output data of the tire model under typical attachment coefficient conditions and fused them into a training data set.The lower layer control used the neural network to perform reverse learning in the form of numerical fit to obtain the reverse tire model.Finally,the desired lateral and vertical control parameters are obtained,and the trajectory tracking control of the vehicle can be adapted to different road conditions.(3)The autonomous collision avoidance mechanism is the guarantee of safe driving.An emergency braking decision module is added on the basis of trajectory tracking control.In order to reduce the unnecessary triggering of the emergency braking system,a dual-scene safety time model is established by dividing the acceleration and deceleration of the target vehicle.Through the model and actual vehicle test,the maximum deceleration that the vehicle can achieve under different road conditions is obtained.The algorithm dynamically updated the safety time model based on the estimated value of the road surface adhesion coefficient.By checking the safety status of the vehicle in real time,the algorithm realized emergency braking control with adaptive road conditions.(4)The reserch also verified the accuracy of the model and the control performance of the algorithm.Under the same control input,the vehicle dynamics model is compared with the Carsim vehicle model and the actual test vehicle parameters.It is proved that the model has a high-precision fit to the dynamic characteristics of the target vehicle.Based on the reliability has been verified vehicle dynamics model,the longitudinal and lateral trajectory tracking control was verified.Under the condition of constant or varying pavement adhesion coefficient with Step change or continuous change target longitudinal speed,it is verified that the longitudinal control strategy has a high accuracy and fast response control effect.Longitudinal control provided stable speed tracking for lateral control.By designing driving scenes with varying road adhesion coefficients,a multi-objective composite test case was established based on the actual collection of complex road trajectories such as park maps.It is proved that the lateral control strategy has strong robustness and high-precision tracking control effect under varying adhesion coefficient road conditions.In the test scenario of China-New Car Assessment Program,the safety collision avoidance performance of the emergency braking strategy under different road conditions was verified.The stable motion control of intelligent vehicles on the road with changing adhesion coefficient is a technical difficulty that unmanned driving needs to break through.The research established a reference vehicle model for variable adhesion coefficient conditions,a pavement adhesion coefficient estimation algorithm with optimal application costs,a lateral and longitudinal trajectory tracking control strategy,a lateral and longitudinal trajectory tracking control strategy that considered changes in pavement adhesion coefficients and different vehicle subsystem characteristics,a emergency collision avoidance mechanism that can adapt to different road conditions.These modules ultimately constituted a complete vehicle motion control system.The ultimate goal of the reserch is to improve the applicability of intelligent driving technology in more environments.