Research on intelligent robot formation based on fuzzy Q-learning

Reinforcement learning is an often used computational approach with simple learning mechanism and needs no environment model. Unlike supervised learning, it has no teacher signal and the decision policy is judged by a reinforcement signal, thus it has a long learning process. In this paper, fuzzy logic and reinforcement learning are combined to improve the learning speed of the formation behavior of the robot. Firstly, a relatively complete database of fuzzy control rules for every behavior is set up with human experience. Then, Q-learning is adopted to adjust the weighting factors of the behavior fusion. Finally, the simulation results are provided to show the validity of the algorithm.