This paper discusses human action modeling and its application to a control system that uses the Mahalanobis-Taguchi System (MTS). In this study, we define embodied knowledge as being included in tacit knowledge. We also define a set of skills based on experiences and intuitive sense as seen in creating an art, sport, craft, or other skilled task. Embodied knowledge is difficult to express explicitly. As our goals, we analyze embodied knowledge acquisition for human action modeling and apply to a control system by using MTS. An analysis of embodied knowledge using devices and pattern-recognition techniques to recognize un-explicit knowledge are being developed owing to recent improvements in technology. Embodied knowledge acquisition element of recognition can be represented as a pattern recognition technique. In this paper, we confirm that MTS is an adaptive method for recognizing pattern in human action modeling. We set up a control model including a controller using the MTS, which is modeled after an internal model of the cerebellum. We apply the controller based on the Recognition Taguchi (RT) method to invert the control of the pendulum. The result indicate that the controller is capable of detecting disturbance.