Controlling a space robot without actuators on the main body is an under-actuated control problem. As an approach to stabilization, various methods such as time-varying feedback controllers, discontinuous feedback controllers, center manifold-based methods, zero-dynamics methods, and sliding mode controllers have been proposed. However, with these methods, modeling errors and time delays have not been sufficiently considered. In order to obtain faster convergence time and compensate for modeling errors and time delays, an adaptive invariant manifold-based switching control method is proposed. In this paper, experiments are carried out to validate the proposed method, using the experimental setup of a planar two-link space robot. The experimental results show that the proposed method is capable of stabilizing the state variables in the required state, even if time delays and modeling errors exist in the system.