Article 4418

Title of the article

METHOD OF SYNTHESIZED OPTIMAL CONTROL FOR A GROUP OF ROBOTS 

Authors

Diveev Askhat Ibragimovich, doctor of technical sciences, chief of the division of safety and nonlinear analysis, Federal research center «Computer science and control» of the Russian Academy of Sciences (Dorodniсyn Computer Center of RAS), professor of the department of Mechanics and mechatronics in RUDN University (119333, 40 Vavilov street, Moscow, Russia), E-mail: aidiveev@mail.ru
Shmalko Elizaveta Yurjevna, candidate of technical sciences, scientific worker of the division of safety and nonlinear analysis, Federal research center «Computer science and control» of the Russian Academy of Sciences (Dorodniсyn Computer Center of RAS), (119333, 40 Vavilov street, Moscow, Russia), E-mail: asiedora@mail.ru 

Index UDK

51-74, 519.6 

DOI

10.21685/2307-4205-2018-4-4

Abstract

Background. A group of robots is a complex control object with dynamic phase constraints. To date, for such objects there are no effective algorithms for solving the problem of optimal control in the initial formulation, when control must be found in the form of a function of time in infinite-dimensional space. The main problem here is that after the reduction of the optimal control problem to the problem of finite-dimensional optimization, we obtain the objective function, which in the parameter space does not have properties of convexity and unimodality.
Matherials and methods. The paper presents a new approach to solving the problem of optimal control – optimal synthesized control. Initially, we provide stabilization of the object with respect to a certain point in the state space, solving the problem of control system synthesis. Then we find such a sequence of stabilization points in the state space that, switching the stabilization points at fixed instants of time, we ensure the movement of the object from the initial state to the terminal state with the optimal value of the quality criterion.
Results. The implementation of the proposed method for solving the problem of optimal control by a group of robots with phase constraints is shown. A comparison of the proposed approach with known methods of finite-dimensional optimization is presented.
Conclusions. It is shown that the gradient classical methods do not find an acceptable solution. Evolutionary algorithms find solutions in all considered cases much better than classical gradient methods and random search algorithm. The method of synthesized optimal control makes it possible to find, with the help of the same evolutionary algorithms, essentially better results than the reduction methods with fewer calculations of the target functional. 

Key words

group of robots, optimal control, method of synthesized optimal control 

Download PDF

 

Дата создания: 04.03.2019 10:35
Дата обновления: 04.03.2019 11:14