For the teleoperation of dual-arm robots with various tasks, the existence of a controller with a high degree of freedom is indispensable. Especially when precise work is required, additional information such as force feedback is very helpful for the operator. In transmitting such force information, a control device of exoskeleton-type with many points of contact with the human body can be one of the solutions. This paper proposes an optimal design method for the 7 degrees of freedom (DOF) exoskeleton systems. The proposed method optimizes the kinematic parameters by using kinematic performance indices related to the dexterity of the human and exoskeleton system. The manipulability ellipsoid is a representative index that can confirm the dexterity of the robot. In this study, we derived the objective function considering the human body model and then optimized it using a genetic algorithm. Unlike other HRI (Human-Robot Interaction) systems, exoskeleton robots share the end-effector as well as the base of the robot with the wearer. Therefore, it is hypothesized that the proposed performance index will be highly suitable for exoskeleton systems.