Research On Trajectory Planning And Coupled Motion Control Of Autonomous Vehicle In Curve Driving

As the future development direction of vehicles,intelligent vehicles have significant advantages in improving traffic efficiency,safety and passenger comfort.With the increasing number of vehicles year by year,traffic congestion,frequent traffic accidents,energy waste and other social problems are becoming more and more serious.Intelligent vehicles have become an urgent need to improve the traffic environment.As a common traffic bottleneck,curve driving not only easily induces road traffic accidents,but also often reduces vehicle traffic efficiency.Compared with human drivers,intelligent vehicles have great advantages in vehicle handling,and have great potential to improve traffic congestion in curves driving.Improving the capacity of intelligent vehicle in curve driving will effectively improve the operation efficiency of public traffic,which is also one of the important issues to be solved in the future development of intelligent transportation.In this paper,the local path planning and vehicle motion control will be studied to improve the capacity of vehicles in curve driving.Firstly,the collaborative analysis of trajectory planning and motion control is carried out based on the optimal preview distance,and the local trajectory planning and traffic speed planning are completed in the rolling window;secondly,the dynamic convergence performance of tracking error is analyzed in the sliding window,then the lateral and longitudinal motion coupling controller is designed to ensure the efficiency and safety of intelligent vehicle in trajectory tracking in curve driving;finally,based on radial basis function neural network,an auxiliary system of error approximation is constructed to further reduce the steady-state error of trajectory tracking,which further optimize the tracking effect in curve driving.This paper focuses on the following aspects.The dynamic and kinematic characteristics of the intelligent vehicle are analyzed,and the preview following kinematics model with preview distance is derived based on Frenet Serret frame.In order to further analyze the influence of road environment and vehicle real-time state on the optimal preview distance,the lateral cascade controller is designed and the joint simulation test under multiple working conditions is carried out.Finally,based on the simulation test data,the neural network mapping model between the optimal preview distance and vehicle speed,initial yaw rate,road curvature,initial lateral deviation and initial heading angle deviation is established.A collaborative analysis method of curve trajectory planning and trajectory following is designed based on optimal preview distance.Based on the sliding window method,a cubic spline curve cluster is generated to meet the constraints of safety and feasibility.Then local optimal trajectory is selected for the vehicle based on a comprehensive performance evaluation function.Based on the sliding window,a curve speed planning algorithm is proposed to generate planning speed,which could guarantee multiple performance demand,such as driving safety,comfort and legitimacy.This paper analyzes and imitates the prediction behavior of human drivers when their vision is blocked in the curve.Based on the Adams extrapolation method,the blind area road in the curve is predicted.Combining the safe braking distance and the preview distance in the curve,a corresponding real-time adjustment strategy of sliding window is established.The results show that the predicted curve extends the longitudinal sight distance and improves the effect of track preview and following.In this paper,the transient response characteristics of tracking error in the process of curve driving are analyzed,and the dynamic constraint of tracking error is transformed into equality constraint by a transformation function;a new convergence function of error constraint is designed to ensure the convergence performance of tracking error in planning window;then,the longitudinal position error,lateral preview error and heading angle deviation are comprehensively considered in the coordinated controller design.The relevant parameters of coordinated controller are designed based on the stability analysis of control system.In addition,the strong coupling of lateral and longitudinal motions of autonomous vehicles also contributes to the saturations damage and increases the control difficulty.To reduce these negative impacts on motion control,a novel motion controller for automated vehicles is presented in this paper.Firstly,a three degree of freedom dynamic model of vehicle is reconstructed with consideration on external disturbance and uncertainties caused by the time varying parameters and unmodeled dynamics.And the saturation constraint of vehicle is put forward by using the hyperbolic tangent function with smooth and approximate saturation characteristics.Then,the error approximation auxiliary system of RBF neural network is designed to optimize and improve the performance of the coordinated controller Parameter design criteria of controller and neural network are given in stability analysis based on Lyapunov Theory.Finally,based on the joint simulation platform of Car Sim Simulink and real vehicle test,the designed trajectory planning algorithm and trajectory tracking controller algorithm is verified,and the test is carried out under different road curvature and different longitudinal speed.The experimental results further verify the feasibility and robustness of the proposed trajectory following motion control algorithm.