A reinforcement learning based robotic navigation system

It is a challenging task for an autonomous robot to navigate in an unknown environment. Machine learning could be useful to support the robot to adapt to the environment and learn the correct navigation skills quickly. In this paper, a reinforcement learning (Q-learning) based approach is proposed to help a robot to move out of an unknown maze. The definitions of the world states, actions and rewards of the algorithm are presented and some experiments are completed to validate the approach. The experimental results show that the proposed approach does have a good performance on mobile robot navigation.