Behavior Control Algorithm for Mobile Robot Based on Q-learning

Abstract

In order to adapt to navigation in unknown environment, the mobile robot must have intelligent abilities, such as environment cognition, behavior decision and learning. The navigation control algorithm is researched based on Q learning method in this paper. Firstly, the corresponding environment state space is divided. The action sets mapping with states are set. And the reward function is designed which combines discrete reward returns and continuous reward. The feasibility of this algorithm is verified by computer simulation.