We introduce hierarchical fuzzy intelligent control for high-level navigation tasks in a unstable system. Control uses linguistic expressed knowledge and consists of three subsystems – planning, navigation, and regulation. System performance is demonstrated in a gymnastic action behavior control problem. Gymnastic action is a difficult task for control, because it has many difficult-to-handle properties such as nonlinearity, numerous degrees of freedom, and unstable behavior. We show the controller realize gymnastic action sequences with linguistic knowledge in 5-link dynamic model computer simulation.