The Application of the Reinforcement Learning Method for Mobile Robot Navigation in an Unknown Environment

Mobile robots have attracted the attention of researchers because of their potential use in industry and daily life. Traditional navigation methods based on the predefined path or known map have been successfully applied to robots working in various scenes. When a robot is operating in an unknown environment, it must learn how to navigate through obstacles, identify risks, and design new trajectories in order to achieve its target. This paper presents the application of the reinforcement learning (RL) method in which the RL algorithm is based on the Q-table for robot navigation. A simulation model is designed on the Gazebo platform for the initial training of RL policies. The simulation and experimental results have proven the proposed method is efficient and the robot works well in an unknown environment involving different obstacles.