In robotics, high mechanical impedance, resolution of sensors, and computing bandwidth have been desirable aspects for feedback control to obtain stability and better performance. In contrast, the human body is physically inferior to man-made systems in terms of feedback control due to low mechanical impedance, resolution of sensors and long signal transmission delay. Nevertheless, humans can achieve not only stable motion generation but also robust and adaptive control ability with respect to uncertain environment. For these reasons, in neurophysiology, many studies have emphasized the significance of the feedforward control ability of human. Virtual trajectory control hypothesis and internal model hypothesis have been referred to explain stable and robust motion generating ability of human under feedforward control. Inspired by the achievement of the previous studies, a hypothesis that joint stiffness can be utilized to compensate inverse dynamics model uncertainty is proposed. To verify this hypothesis, a robot which can control joint stiffness is developed. From the experimental results, the developed robot shows stable and accurate motion tracking performance without any sensors and precise inverse dynamics model. These results indicate that the concept of the developed variable stiffness actuator can be applied to the robotics field for feedforward application, and support the hypothesis that human utilize joint stiffness to adapt to uncertain environment.