Issues of adaptive fuzzy direct compensation-based state/output feedback control for nonstrict-feedback systems are presented. The key to its feasibility is the differentiation-free feature, which is achieved in two steps. First, with the nominal adaptive fuzzy virtual controllers as the inputs, a set of low-pass filters are constructed to avoid the explosion of complexity and the algebraic-loop problems. Second, via using sufficiently small time constants, the boundedness of the filters' errors is ensured without the calculation of the filter error dynamics, which otherwise would incur another loop problem. In particular, by including a supervisory linear high-gain control component, both the state/output feedback control schemes ensure the semi-global practical tracking stability without relying on the all-time validity of the fuzzy approximation. In particular, the stability criteria of the proposed state and output feedback designs are much easier to fulfill than those based on the variable-separation-based method. Simulation are then carried out to validate the proposed schemes.
