Machine Learning of a Universal Motion Stabilization System in the Advanced Optimal Control Problem

Abstract

This paper discusses the advanced task of optimal control. In this task, it is necessary not only to find a control function as a function of time to achieve a control goal with an optimal value of a given quality criterion, but also to ensure the existence of an attraction property of an optimal trajectory in some of its vicinity. This property allows you to find a control function that can be implemented directly in a real object. To solve the problem, a universal system for stabilizing the movement of a control object along a given trajectory is first synthesized. Here machine learning of control by symbolic regression is used. Symbolic regression allows you to find the structure and parameters of the control function without human participation. To ensure the versatility of the synthesized stabilization system, symbol regression looks for a single stabilization system for some given of different trajectories. The synthesized stabilization system and the initial mathematical model of the control object are entered into the object control system. The obtained mathematical model of the control object allows solving an extended problem of optimal control and obtaining a control function implemented directly in a real object. Presented is a method for solving an extended problem of optimal control of spatial motion of a quadcopter

The work was partly supported by the Russian Science Foundation (project no. 23-29-00339)

Please cite this article in English as:

Diveev A.I. Machine learning of a universal motion stabilization system in the advanced optimal control problem. Herald of the Bauman Moscow State Technical University, Series Instrument Engineering, 2025, no. 1 (150), pp. 71--90 (in Russ.). EDN: XKJFLR

References

[1] Mitchell T.M. Machine learning. McGraw-Hill, 1997.

[2] Pontryagin L.S., Boltyanskiy V.G., Gamkrelidze R.V., et al. Matematicheskaya teoriya optimalnykh protsessov [Mathematical theory of optimal processes]. Moscow, Nauka Publ., 1983.

[3] Shen X., Shi W. Adaptive trajectory tracking control of wheeled mobile robot. Proc. CCDC, 2019, pp. 5161--5165. DOI: https://doi.org/10.1109/CCDC.2019.8833019

[4] Samir A., Hammad A., Hafez A., et al. Quadcopter trajectory tracking control using state-feedback control with integral action. Int. J. Comput. Appl., 2017, vol. 168, no. 9, pp. 1--7.

[5] Uddin N. A robot trajectory tracking control system design using pole domination approach. IEEE IEMCON, pp. 506--512. DOI: https://doi.org/10.1109/IEMCON.2018.8614899

[6] Andreev A., Peregudova O., Sutyrkina K., et al. On global output feedback trajectory tracking control of a wheeled mobile robot. Proc. ICMT, 2019. DOI: https://doi.org/10.1109/ICMECT.2019.8932122

[7] Su T., Liang X., He G., et al. Robust trajectory tracking of delta parallel robot using sliding mode control. IEEE SSCI, pp. 508--512. DOI: https://doi.org/10.1109/SSCI44817.2019.9003125

[8] Kolesnikov A.A., Kolesnikov Al.A., Kuzmenko A.A. The ADAR method and theory of optimal control in the problems of synthesis of nonlinear control systems. Mekhatronika, avtomatizatsiya, upravlenie, 2016, vol. 17, no. 10, pp. 657--669 (in Russ.). DOI: https://doi.org/10.17587/mau.17.657-669

[9] Kolesnikov A.A., Kolesnikov Al.A., Kuzmenko A.A. Backstepping and ADAR method in the problems of synthesis of the nonlinear control systems. Mekhatronika, avtomatizatsiya, upravlenie, 2016, vol. 17, no. 7, pp. 435--445 (in Russ.). DOI: https://doi.org/10.17587/mau.17.435-445

[10] Nguyen A.T., Xuan-Mung N., Hong S-K. Quadcopter adaptive trajectory tracking control: a new approach via backstepping technique. Appl. Sci., 2019, vol. 9, iss. 18, art. 3873. DOI: https://doi.org/10.3390/app9183873

[11] Diveev A.I., Sofronova E.A. Universal stabilisation system for control object motion along the optimal trajectory. Mathematics, 2023, vol. 11, iss. 16, art. 3556. DOI: https://doi.org/10.3390/math11163556

[12] Diveev A.I., Shmalko E.Yu. Machine learning control by symbolic regression. Cham, Springer, 2021. DOI: https://doi.org/10.1007/978-3-030-83213-1

[13] Karpenko A.P. Sovremennye algoritmy poiskovoy optimizatsii. Algoritmy, vdokhnovlennye prirodoy [Modern search engine optimization algorithms. Algorithms inspired by nature]. Moscow, BMSTU Publ., 2014.

[14] Simon D. Evolutionary optimization algorithms. Wiley, 2013.

[15] Diveev A.I., Konstantinov S.V. Study of the practical convergence of evolutionary algorithms for the optimal program control of a wheeled robot. J. Comput. Syst. Sci. Int., 2018, vol. 57, no. 4, pp. 561--580. DOI: https://doi.org/10.1134/S106423071804007X

[16] Diveev A.I., Sofronova E.A., Konstantinov S.V. Approaches to numerical solution of optimal control problem using evolutionary computations. App. Sci., 2021, vol. 11, iss. 15, art. 7096. DOI: http://dx.doi.org/10.3390/app11157096

[17] Diveev A. Hybrid evolutionary algorithm for optimal control problem. In: Arai K. (ed). Intelligent Systems and Applications. IntelliSys 2022. Lecture Notes in Networks and Systems, vol. 543. Cham, Springer, 2022, pp. 726--738. DOI: https://doi.org/10.1007/978-3-031-16078-3_50

[18] Davis L. Handbook of genetic algorithms. Van Nostrand Reinhold, 1991.

[19] Kennedy J., Eberhart R. Particle swarm optimization. IEEE ICNN, 1995, vol. 4, pp. 1942--1948. DOI: https://doi.org/10.1109/ICNN.1995.488968

[20] Mirjalili S., Mirjalili S.M., Lewis A. Grey wolf optimizer. Adv. Eng. Softw., 2014, vol. 69, pp. 46--61. DOI: https://doi.org/10.1016/j.advengsoft.2013.12.007