Development of a Human-Like Adaptation Rule-based Proportional-Integral Control Algorithm for Universal Steering Control of Autonomous Mobility with RLS

For autonomous mobility control, the need for a controller that guarantees robust performance for path tracking as well as enhancement of user convenience is emerging. This paper describes a human-like adaptive rule-based Proportional-Integral(PI) control algorithm for universal steering control of autonomous mobility using recursive least squares with forgetting. In this study, adaptation rules that represent human adaptation characteristics has been designed by relation functions designed using time derivatives of control error and input. The sensitivity in the function was estimated using the recursive least squares method. The integral input was derived using the Lyapunov direct method and the gradient descent method. The proportional input was derived based on the gradient descent method using the sensitivity of the time derivative of control error with respect to the time derivative of control input. The performance of the proposed control algorithm was evaluated under Matlab/Simulink and CarMaker environments. As a result of performance evaluation, the human-like adaptation rule-based control algorithm proposed in this study shows reasonable target value tracking performance without system information like dynamics and parameters.