Self-organizing map for reinforcement learning: obstacle-avoidance with Khepera

摘要

We present a self-organizing map implementation of the Q-learning algorithm. Our goal is to overcome the problems of reinforcement learning: memory requirement and generalization. We consider the map as an associative memory and we use it for obstacle avoidance with the mobile robot Khepera. Results allow real world applications to be envisaged using neural reinforcement learning.