Research on adaptive heuristic critic algorithms and its applications

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 is used to learn the actions in interaction with the environment in designing intelligent robot. In this paper, the actions that 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. Adaptive heuristic critic are dominant learning algorithms of reinforcement learning method. First, the implement method of adaptive heuristic critic by neural networks are presented in this paper and then collision avoidance behavior learning of intelligent robot has been settled by use of these algorithms. From the test results given in paper, it is easy to see that robot can learn the collision avoidance behavior by itself and its adaptation to environment has been improved enormously.