In modern realities, the principle of adaptability becomes an absolute necessity for the normal functioning of complex technical systems. To achieve adaptability, controller synthesis can be based both on the classical theory of automatic control and using various approximate methods of intelligent control. This paper analyzes publications from 2014 to 2024 on new approaches to the design of adaptive control systems for various moving objects with a drive actuator. The first part of the review is devoted to classical methods, including the adaptive controller with a reference model, and its new areas of application in technology (control of a vibration machine and a Stuart platform). The similarities between classical adaptive control and machine learning are noted. The second part presents the results of research based on the joint use of a classical controller and various intelligent methods, such as fuzzy logic, neural networks and machine learning, forming complex multi-component control structures. The results show that the use of such an integrated approach can significantly improve the performance of the main controller, expanding its adaptive capabilities with respect to uncertainties and parameter changes, disturbances and the effects of nonlinearities.

This paper presents the problem of adaptive output tracking for a class of unstable multi-input multi-output linear time-invariant systems affected by unknown external disturbances, taking into account different control delays across the channels. It is assumed that both the reference signal and the external disturbances have a multi-harmonic form with unknown frequencies, phases, biases, and amplitudes. The disturbances may be unmatched in the system and affect both the inputs and the outputs. For such systems, a linear state feedback control law is first designed based on the classical Falb—Wolovich method to decouple the multivariable system into independent control channels. This approach to the channel decoupling allows, in the case of the presence of nonminimum-phase zeros in the transfer function, to exclude them and transform the original system into transfer functions with independent integrators. Then, an observer is constructed to estimate the states of the reference signal and the external disturbances. Finally, an adaptive controller is synthesized to compensate for the external disturbances and ensure accurate output tracking of the reference signal. In this work, the adaptive algorithm with memory regressor extension is employed with the aim of improving the convergence rate of the parameter adaptation. The proposed method guarantees that all signals in the closed-loop system remain bounded and ensures the asymptotic stability of the output. To validate the proposed approach, numerical simulations are carried out in the MATLAB environment. The proposed solution is feasible for "square" systems, when the number of inputs and outputs of a multi-channel system is the same.

The paper considers the problem of optimal tracking control for nonlinear systems on a finite time interval. In this case, the system is represented in the state space form with state-dependent coefficients (SDC) matrices. The problem of finding a solution to the tracking problem on a finite time interval in the nonlinear SDC formulation is associated with finding a solution to the state-dependent differential Riccati equation and the differential equation for the auxiliary feedforward vector, the initial conditions for which are usually specified at the right end. А typical approach to solving such problems uses the integration of these equations in the backward direction (from right to left), where the calculation of the SDC matrices of the system requires information on the state variables of the system and control, which is not available without additional measures. To overcome the indicated problem of the unknown state vector during backward integration, this paper proposes an approach based on deriving a solution through the corresponding differential equations for the Riccati matrix and the auxiliary vector, the initial conditions for which are uniquely specified at the left end of the time interval due to the use of a special Riccati transform, different from the typical one. This allows calculating the control through the integration of the corresponding differential equations in direct time, which removes the problem of the unknown state vector. The proposed approach is tested on the academic example of the Van der Pol oscillator, for which an additional study of the effectiveness of the proposed method in comparison with the most popular existing approaches is performed. The results of computer modeling confirmed the advantage of the proposed method, both in terms of the terminal accuracy of tracking the driving signal, and in terms of the mean square error of tracking.

A significance of quadruped robots lies in their unique design and functionality, which allow them navigating a variety of terrains with remarkable agility and stability. Unlike popular wheeled robots, quadrupeds mimic locomotion of four-legged animals, enabling to traverse uneven surfaces, climb obstacles, and maintain balance in challenging environments. This paper introduces a development of a new leg-wheel hybrid quadruped robot. А peculiar design of the quadruped robot allows the robot to function as a bipedal robot while performing stationary tasks that do not require to change its location in space; at the same time, wheels at legs’ endpoints allow fast locomotion on a flat rigid terrain. The dual functionality enhances its versatility and broadens a range of tasks it can perform, making it suitable for various applications in research and practical applications. Design procedures, modelling methodologies, and static structural analyses performed to finalize a structure of the robot are demonstrated in the paper.

Knowledge of the manipulator’s joints and links stiffness parameters is often necessary for high-precision control in the presence of external loads. The article proposes a method for identifying torsion stiffness coefficient of a rotational joint of a 1-DOF manipulator using a microelectromechanical inertial measurement module (IMU) consisting of an accelerometer and an angular velocity sensor. The method is based on describing the dynamics of a joint with an unknown torsional stiffness coefficient using a transfer function (TF) with known input and output signals (the angle of rotation of the motor and the angle of rotation of the link, respectively). The input signal of the TF is read from the engine encoder, and the output signal is reconstructed from the data from the IMU. When harmonic signals with different frequencies are assigned to the drive, the amplitude change and phase shift of the signal passing through the system are measured. Based on the data obtained, the amplitude-phase frequency response of the system is constructed, from which the time constants of the TF describing this system are calculated. The time constants depend on the mechanical characteristics of the system, including the torsion stiffness coefficient of the virtual spring, which is calculated based on the identified time constants. The method, unlike the existing ones, does not require bulky equipment for applying external forces to the manipulator, nor expensive measuring systems for measuring displacements in space. The results of experimental validation on a 1-DOF manipulator confirm the workability of the proposed method.

The article discusses the solution of the scientific problem of ensuring stable orientation of a weakly asymmetric spacecraft in the atmosphere of Mars. The solution of the problem is required to perform the failure-free deployment of braking parachute. The known quasi-static equations of spacecraft rotational motion has a small mass-aerodynamic asymmetry are applied. Let us noted that before solving the control synthesis problem, these equations are linearized by two orientation angles are three angular velocities. The aim of the work is to ensure a law of optimal controlled stabilization of a spacecraft with small mass-aerodynamic asymmetry. Simultaneous stabilization is achieved in terms of orientation angles and angular velocity vector during descent into the Martian atmosphere. When synthesizing expressions for optimal stabilization of a spacecraft, the classical method of dynamic programming is used. The main simplifying assumptions used in the work are: assumptions about the smallness of angular velocities, angles of attack and slip, displacement of the center of mass, aerodynamic moments from the violation of the axisymmetric shape, as well as the assumption about the implementation of a coplanar combination of the asymmetry of the spacecraft. Based on the results of numerical modeling: the use of the obtained expressions for control allows minimizing the orientation angles and angular velocities to the required small values over a time interval equal to a half of the movement interval from the beginning of the atmospheric descent to the moment of opening of the braking parachute.