Describir: Hierarchical Reinforcement Learning Using a Modular Fuzzy Model for Multi-Agent Problem