Employing of RL Technology to Develop an Adaptive Motion Controller for a Line Follower Robot

Abstract

The article is focused on the development process of an adaptive motion controller for a line follower robot. The controller learning process took place on the basis of the digital twin of the mobile robot using reinforcement learning technology. The digital twin and the reinforcement learning algorithm were implemented in MATLAB/Simulink. The Twin–Delayed Deep Deterministic Policy Gradient Agents method was used as a learning algorithm. The reward function was taken to minimize the distance between the center of the robot and the middle of the nearest section of the color–contrast line, as well as the difference between the angle of the robot position and the tangent to the current section of the line.