This paper addresses an autonomous space robot for a truss structure assembly using some reinforcement learning. It is difficult for a space robot to achieve a task on contact with the real environment, because of the error between the real environment and the model of controller. In order to solve the problem, we propose an approach that a space robot autonomously obtains proficient and robust skills against error of the model to achieve a task. Numerical simulations and experiments show that the proposed method is useful in the real environment.