Q-LEARNING AND ITS APPLICATION IN LOCAL PATH PLANNING OF INTELLIGENT ROBOTS

Abstract

The concept of reinforcement learning comes from behavior psychology that takes behavior learning as trial and error, by which the states of environment are mapped into corresponding actions. There's a question of how the behaviorism can be used to learn the actions in interaction with the environment in designing intelligent robots. In this paper, the actions that a robot takes to avoid obstacles are taken as one class of behaviors and the reinforcement learning is used to realize behavior learning of obstacle avoidance. Q\|learning is one kind of reinforcement learning method that is similar to dynamic programming. After basic ideas of Q\|learning are introduced, a neural network learning algorithm of Q\|learning with concepts of competition and self\|organization is presented. Its application in local path planning of intelligent robots is also introduced. Finally, the detailed simulation results are presented.