The Reaserch On Path Tracking Control With Delay Conpensation For Autonomous Vehicles

In recent years,the increase in car ownership has accelerated the circulation of goods and people,and improved people’s material living standards,at the same time,impacted in traffic safety,traffic congestion,environmental pollution and energy consumption.Autonomous driving technology has the capability in improving safety,traffic efficiency,energy consumption.Path tracking control is one of the key technologies of autonomous driving.Due to the time taken for sensor data transmission,controller data processing,and actuator steering actions,delay phenomenon are unavoidable in autonomous vehicle control systems.The existence of the steering delay phenomenon will deteriorate the performance of the path tracking controller.In response to this problem,by applying the method of simulation verification platform of the Carsim and the mobile platform test car,this thesis studied the path tracking control method of autonomous driving vehicles considering delay compensation.The details are as follows:(1)When the actuator was in an ideal state without delay,the path tracking control algorithms were analyzed and compared: Stanley without preview,Pure Pursuit(PP)with single-point preview,and Model Predictive Control(MPC)with multi-point preview and its variant-Spatial Model Predictive Control(SMPC).The selection of controller parameters has an important affect on its performance.In order to avoid the subjectivity of manual parameter adjustment,the genetic algorithm was used to optimize the controller parameters.Among them,MPC can systematically consider the control problem with constraints,can control the multi-input multi-output system by modeling,and has scalability and can be applied to the motion planning problem,which is the focus of research.SMPC transforms the kinematic deviation model in the time domain used by MPC into the space domain,thereby formulating an objective function that minimizes the first-order and second-order change rates of steering curvature,and outputs a smoother steering control quantity.Through Carsim/Simulink co-simulation,the results showed that all four controllers could control the vehicle for path tracking with low lateral deviation.(2)The steering delay model was established to simulate the actual actuator with delay phenomenon,and the path tracking control algorithms were analyzed and compared:Stanley,PP,SMPC.The simulation results showed that the delay would cause the control variables of the three controllers oscillate,the vehicle trajectory would appear "wave-shaped",and the performance of Stanley without preview would deteriorate more seriously.Based on SMPC,a prediction strategy for compensation in the output phase of the control quantity,a prediction strategy for compensation in the vehicle state acquisition phase,and an augmentation strategy for compensation by considering the delay model in the building of the vehicle motion deviation model were proposed.The simulation results showed that two prediction strategies and the augmentation strategy both could improve the performance of the controller in case of steering delay,and the augmentation strategy was better than the prediction strategies.(3)The experimental research was carried out through the mobile platform test car equipped with the combined inertial navigation system and the Speedgoat mobile real-time target machine.In order to keep the vehicle running at a constant speed,the driving reaction of the car was formulated through a given throttle test on a straight road and a PID controller was designed.When the actuator had no delay,the path tracking tests of PP,Stanley,MPC and SMPC were carried out.The test results showed that the four path tracking controllers could control the vehicle to track the reference path with low lateral deviation,and the PID controller could keep the vehicle running at a constant speed with low speed deviation.When the actuator had delay and the path tracking controller was SMPC,the effect verification test of the delay compensation strategy and the test of the influence of parameters on the performance of SMPC and compensation strategy were carried out.The experimental results showed that two prediction strategies and the augmentation strategy both could improve the performance of the controller in case of steering delay,and the augmentation strategy was better than the prediction strategies;a smaller pure lag time constant could reduce the impact of the steering delay on SMPC,Smaller discrete time steps and prediction time domain would exacerbate the impact of steering delay on SMPC,but augmentation strategies could invariably improve the performance of SMPC.