Stabilisation System Synthesis for Motion along the Trajectory by Evolutionary Machine Learning Control

Askhat Diveev1,Email

Elena Sofronova1,Email

Nurbek Konyrbaev2,Email

Sabit Ibadulla2

1Department of Robotics Control, Federal Research Center “Computer Science and Control” of the Russian Academy of Sciences, 119333, Vavilova Str., 44, Build. 2, Moscow, Russia.
2Department of Computer Science, Institute of Engineering and Technology, Korkyt Ata Kyzylorda University, 120014, Aiteke bi Str., 29A, Kyzylorda, Kazakhstan.

Abstract

The article discusses the problem of synthesising a system for stabilising the movement of an object along a given trajectory. Solving the control synthesis problem involves finding the control function on the deviation of the object from a given trajectory. A trajectory stabilisation system is necessary for the object to maintain its trajectory under real conditions in the presence of external disturbances. In the work, machine learning control by symbolic regression was used to solve the control function synthesis problem. Symbolic regression methods allow to find the mathematical expressions of the desired functions in the form of special code. To find the mathematical expression of the desired function, the symbolic regression method uses a special genetic algorithm that searches the code space for the optimal solution according to the given optimisation criterion. An example of motion stabilisation of two quadcopters along optimal trajectories is presented.

Stabilisation System Synthesis for Motion along the Trajectory by Evolutionary Machine Learning Control