Miniaturization and improved performance of computers, sensors, and executive devices open up new possibilities for intelligent control over complex mechatronic systems on transition processes and under turbulence. Previously, adaptive control problems in a changing environment and with state space structure changing in time have virtually not been studied due to limited capabilities for a practical implementation. The change of the state space structure (dimension) is possible under active interaction with the environment. In this work, with the example of an airplane with a large number of "feathers" distributed on the surface, i.e., elements with pressure sensors and executive turning devices, we show that in complex systems adaptation to changes in the structure of external disturbances can be done with internal selforganization, similar to multiagent systems. Under turbulence, to solve the problem of equating the influence of disturbing forces on different elements of the wing and transforming airplane flow to a mode close to laminar, we have considered the possibility to apply a multiagent protocol. The operability of the new approach is illustrated with a specially made experimental testbed that allows for real physical experiments.