Obstacle Avoidance Research For A Wheeled Mobile Robot Based On Reinforcement Learning

In recent years, the application of robot is more and more extensive, but the actual environment is dynamic, and full of uncertain. If the application of robot wants to reach the all respects of social, the robot has higher intelligence. The contribution of this thesis is the simulation and testing of wheeled mobile robot avoiding obstacles. The main contents and related original achievements are as follows:The current situation of both the domestic and foreign robot control technology is analyzed in detail. The control strategy of mobile robot based on reinforcement learning is proposed. BP neural network is combined to Q learning algorithm. The value of value of reinforcement function is conformed by artificial potential field. The obstacle avoidance is realized though the above-mentioned method, and the mobile robot is controlled to reach goal finally.Secondly, one kind of wheeled mobile robot is designed, with MCS-51 single chip as the master controller. The entire process of main circuit design, programming, com-piling and debugging has been done. Then we apply the program of avoiding obstacles to the robot and let the robot avoid obstacles.In order to verify the effectiveness of these propose methods, this paper finally gives the results of all kinds of obstacle avoidance simulation in a computer virtual environment with Visual C++ and also presents wealthy real experiments indoors, and result of it is analyzed. The result indicates this method can realize the mobile robot avoiding the obstacle.The research of avoiding obstacles behavior has significant meaning for wheeled mobile robot to be autonomous and intelligent.