This article presents a collaboration of a behavior-based control and fuzzy q-learning for a mobile robot navigation system. Many Fuzzy Q-learning algorithms have been proposed to yield an individual behavior like obstacle avoidance, finding a target and so on. However, for complicated tasks, it is needed to combine all behaviors in one control scheme. Based on this fact, this paper proposes a control system that considers and includes fuzzy q-learning in a behavior-based control to overcome complicated tasks in navigation systems of the autonomous mobile robot.
There are two behaviors trained using fuzzy q-learning while other behaviors are coded by hand. Furthermore, a hierarchical hybrid coordination node is used to coordinates all behaviors. Simulation results demonstrate that the robot with proposed one is able to learn the right policy, which is to avoid the obstacle and to find the target. However, Fuzzy q-learning failed to produce the right policy for the robot to avoid the collision in the corner location.