Image base visual servoing base on reinforcement learning for robot arms

摘要

In this paper, the main objective is to design a gain controller for a robot arm based on reinforcement learning methods. The controller is applied in image base visual servoing. The image feature error is used to form the state space. When the robot perceives a state by using a camera, the controller based on Q-Learning will output a control gain to the robot arm. Because Q-Learning does not have any knowledge about the environment, it is suitable to be controller for decision making. After some learning iterations, the controller can output a series of control gain to achieve the goal. The results of simulation confirm the proposed method that can achieve the goal of sample efficiency.