LASSO Regression and Fictitious Reference Signal for Non-Parametric Controller of Mobile Robot

When the structure of the ideal controller is unknown, it is logical to implement an FIR-type controller. Here, we use FRIT for the case of an FIR-type inverse controller. Generally, FIR might have an unnecessarily large number of taps. To counteract this, we employ LASSO regression to achieve an FIR-type inverse controller with as few taps as possible. Additionally, we extend this method to the yaw-rate control of mobile robots. The effectiveness of the proposed method is confirmed using a nonlinear dynamical bicycle model.