Design Of Model Predictive Trajectory Tracking Control Algorithm By Considering Tire Dynamic Constraints

With the rapid development of science and artificial intelligence technology,driverless vehicle technology has received extensive attention,and trajectory tracking control as part of driverless vehicle technology is the focus of research.Since the dynamic characteristics of tires seriously affect the handling stability of vehicles,this paper studies the trajectory tracking control problem of driverless vehicles based on model predictive control by considering the nonlinear dynamic characteristics of tires.The main research of this paper is as follows:(1)The modeling process of vehicle trajectory tracking model and vehicle pre-sight model is described,then the tire lateral force estimator based on radial based(RBF)neural network is established by considering the influence of tire dynamics on vehicle handling stability.(2)For the established vehicle trajectory tracking model and vehicle presight model,the zero-order holding(ZOH)method is used to discretize the continuous equation of state and use it for the state prediction of the trajectory tracking system;In solving the optimization problem,the objective function consists of the slip constraint term,the vehicle trajectory tracking term,and the control input term,and the boundary condition of the slip constraint is determined by the lateral force of the tire estimated by the RBF neural network.Then,the desired longitudinal acceleration and the desired front wheel steering angle are calculated based on the optimization problem.Finally,the vehicle is controlled to achieve trajectory tracking through the lower controller converted into vehicle steering angle,throttle opening and brake master cylinder pressure.(3)In order to verify the effectiveness of the proposed model prediction trajectory tracking control algorithm,this paper conducts the co-simulation test and a hardware-in-loop test of the proposed trajectory tracking control system.A co-simulation platform based on Matlab/Simulink and CarSim is built,and the simulation results show that the effectiveness of the RBF neural network tire lateral force estimator designed in this paper effectively reduces the slip phenomenon of vehicles.The vehicle can still drive effectively beyond the nonlinear area of the tires;Compared with the algorithm proposed in recent years,the algorithm proposed in this paper can effectively drive on the double-line trajectory at a higher speed and has higher tracking accuracy.In order to further verify the engineering feasibility of the algorithm,this paper also performs hardware-in-loop testing of the algorithm.Hardware-in-loop testing was conducted using the Raspberry Pi and Car Sim.The hardware-in-loop test results show that the vehicle can drive safely under extreme working conditions and effectively reduce the slip phenomenon.