Stability Analysis For Autonomous Driving Based On Model Predictive Control

As the new stage in the development of transportation industry,it is of great importance to conduct safety research on autonomous driving.Autonomous driving includes control,perception,planning and other technical aspects,autonomous driving control which as the underlying implementation,is the basis for realizing automatic driving.The good control or not of the implementation control layer directly affects the stability of the vehicle.In this paper,we focus on the stability of vehicle path tracking control,proposing an optimized controller based on model prediction algorithm and conducting stability analysis to complete the safety research on autonomous driving vehicles.At the beginning,this paper carries out vehicle lateral dynamics modeling,deriving the dynamics equations of motion characteristics and attitude position and discretizes the derived equations.Then introduces the basic principles of model predictive control,listing the discrete state space model after augmentation,building the control model on the joint simulation platform of Carsim/Simulink,and provides the subsequent control strategy design and controller improvement.Proposing an adaptive gain control strategy which is based on MPC.The weight ratios of the vehicle yaw angle and trajectory position are taken as the target objects,the adaptive control strategy changing the input weights with the speed is designed after simulating each set of weight ratios at different speeds.The adaptive gain MPC controller combined with this control strategy is in comparison with the conventional preview control so as to verify the effectiveness of the proposed control strategy whether improving the vehicle stability performance.At last,the LQR controller was improved by using a reference path planning algorithm different from the one used in the MPC controller and adjusting the control structure.The improved LQR controller was compared with the MPC controller under speed and road adhesion coefficient conditions,the improved LQR controller showed better performance in controlling the vehicle driving,and the response parameters reflect the stable driving condition of the vehicle.The simulation data are analyzing to establish the safety envelope within the yaw rate –sideslip phase diagram,which can be regarded as a reference criterion for future vehicle safety research.The future research will focus on novel driving control combined perception layer.