We have considered the problem of adaptive suppression of friction self-oscillations in a robot-tool-working surface system. These oscillations appear under contact interaction of the tool with the surface in case of nonlinear friction in the contact area and creeping speed of the tool motion along the surface. Frictional self-oscillation suppression is carried out using an external pulse load on the system. The friction model with Stribeck’s effect in the contact area, inertia and flexibility of the system elements are taken into account in modeling, making it possible to obtain processes close to those observed in practice. It is shown that friction self-oscillations with short-term tool stops arise under these conditions at slow speed of the tool motion along the surface. The pulse load on the system smoothes the processes but self-oscillations appear again after unloading of the pulses. In this paper, we propose an adaptive version of pulsed smoothing. Pulses are fed to the system when auto-oscillations occur. A small increase of the prescribed speed of the tool motion after pulse unloading eliminates the conditions in which self-oscillations occur. The proposed algorithm is implemented using the logical block whose efficiency is confirmed by modeling in Matlab.