Robust control of non-deterministic object (NO) is an urgent task, and there are different approaches to its solution. The paper considers an introduction of fuzzy feedback in the control loop of NO. The value of the feedback coefficient may range from logic zero (the gap feedback) to logic one. Process of increasing the oscillation in the control loop (NO) is controlled by the frequency of oscillation index. The numerical value of the frequency of oscillation index sets parameters of a "forbidden" circle, which includes the "dangerous" point with coordinates -1, j0. Changes in frequency response of NO will be countered by changing the parameters of a "forbidden" circle. In the control loop of the NO we introduce an adaptive nonlinear approximator (regulator). The controller is implemented on the basis of an adaptive neural fuzzificator and unit activation functions block. Activation function (sigmoidal unbalanced function) strengthen "weak" and weakening "strong" signals at the input. Adaptation of neural fuzzificator performed using neuron feedback with sequential learning. Control algorithm feedback coefficient is made on the basis of multilevel relay element (MRE). MRE controlled by duty cycle, proportional to the difference between the signals of the current and minimum values of the frequency of oscillation index. In the control loop of MO desired dynamics is provided by the adaptive fuzzy control, and stability - by fuzzy feedback. The main conclusions are the following: adaptive fuzzy controller to control the MO using neural technology is developed; control algorithm of fuzzy feedback control loop of NO is developed.