Controller Tuning Based on Data-Driven Prediction of Vehicle by Using the Initial Experiment Data in the case where Rate of Control signal is Saturated

Abstract

Estimated response iterative tuning (ERIT) can updated the feedforward controller so as to realize desired closed-loop responses. If ERIT can be applied to vehicle yaw-rate control of automated vehicle, good autonomous driving can be realized. However, ERIT cannot realize desired control performance in the case where the control signal velocity is saturated. To solve this problem, I propose a new data-driven tuning method for vehicle yaw-rate control. Proposed method can update the feedforward controller by using estimation method of a closed-loop transfer function. Compared to ERIT, the proposed method can realize desired control performance even when the initial steering angle velocity is saturated. Furthermore, the proposed method can predict vehicle yaw-rate/steering angle velocity responses. The validity of proposed method is verified through vehicle simulators.