The arrival of an aging society brings up many challenges, including the demanding need in medical resources. In responding, the exoskeleton robot becomes one of the focuses, which provides assistance for people with locomotive problems. Motivated by it, our laboratory has developed a wearable upper-limb exoskeleton robot, named as HAMEXO. It is of 2 DOF and intended to provide motion assistance for users in their daily activities. To serve the purpose, HAMEXO is equipped with a visual system to detect objects in the environment, and also a motion controller for its governing. To deal with the coupling involved during the movements of the two joints and the need to adapt to various users, we adopted the learning approach for controller design. Experiments are performed to demonstrate its effectiveness.