IZVESTIYA SFedU. ENGINEERING SCIENCES

One of the priority areas of group control theory for the near future is swarm control of groups of
small unmanned aerial vehicles - micro-, mini- and nano-classes, performing a collective task under enemy
influence. Here, two antagonistic strategies collide - minimization of losses from the point of view of
the attacking swarm and maximization of such losses from the point of view of the defense system. Research objective: development of an approach to solving a practical problem - penetration of a swarm of
unmanned aerial vehicles into an object protected by a defense system. The objectives of the study were to
analyze the characteristics of the factors influencing the processes of detection, tracking, recognition of
swarm intentions by the defense system and the development of a machine learning model for creating
spatio-temporal formations that minimize the number of swarm elements affected by the defense system.
The main parameters of the defense system are the detection range and duration of swarm recognition, the
time to make a decision on the actions of the swarm, the size of the zone of destruction of defense means.
The method of machine learning on convolutional neural networks with reinforcement was chosen as the
research method. The counteraction effect against the defense system is created due to the swarm's dynamics;
it can actively maneuver, creating spatio-temporal maneuvers during the mission. To simulate the
"Swarm vs. Defense System" situation, a swarm agent (a neural network with a transformer architecture
that initiates swarm formations) and a defense system agent are introduced that recognizes the swarm and
attacks it, creating a zone of destruction in the conventional center of mass of the swarm. The swarm is
guided by a stochastic rule, asking the defense system (environment) to react to its maneuver. The environment
responds by attacking the swarm, creating a damaging factor at the point where the swarm or the
main part of the swarm is expected to be. The reward of the swarm strategy is the number of undestroyed
objects under the conditions of constraints; for the defense system, this "reward" acts as a "punishment".
An interesting phenomenon was established in the process of machine learning: each swarm element,
remaining within a given space and implementing the biological principles of swarm control without a
Leader, independently evades the area of destruction, which together creates a random spatio-temporal
formation for defense means with minimal losses of swarm elements. Thus, using the method of machine
learning with reinforcement, a model was created that allows varying the behavior of the swarm and synthesizing
spatio-temporal formations that complicate detection, tracking, recognition of intentions and
decision-making on the impact of the defense system on a swarm of attacking small unmanned aerial vehicles,
as well as significantly reducing their losses.

In this paper, we investigate the application of neural network architectures for semantic segmentation
of rodent burrows for monitoring their population in agricultural fields. In particular, three models
for semantic segmentation are considered: convolutional autoencoder (CAE), SegNet, and U-Net. These
models are applied to analyze images obtained from unmanned aerial vehicles (UAVs) and ground robotic
means, which allows for automatic burrow detection, minimizing the need for labor costs in processing
large amounts of data. A sample of 247 RGB images containing 1098 labeled burrows was prepared for
training and testing the models. The quality indicators of semantic segmentation were assessed using the
Jaccard metric (IoU), which resulted in the following values: 0.511 for CAE, 0.548 for SegNet, and 0.529
for U-Net. An assessment of the computational resources required to implement these models in on-board
computing units (OCUs) of mobile robotic means was conducted. Two criteria were considered: the number
of floating-point operations (GFLOPS) and the number of model parameters. The results showed that
SegNet requires 2.23 GFLOPS and has 0.76 million parameters, which is 2.58 and 2.33 times less than
SAE and U-Net, respectively. The number of floating-point operations for SegNet was also 2.43 and 1.88
times lower than that of SAE and U-Net, respectively. As a result, SegNet outperformed SAE and U-Net in
both segmentation efficiency and required computational resources. This work was carried out as part of
the implementation of a computer vision system for an agricultural robotic means.

The article presents the results of the development of nanoscale memristive structures, the application
of which is promising for the hardware implementation of artificial intelligence systems. A design of a
memristive cell based on a titanium oxide film with a thickness ranging from 3 to 50 nm is proposed.
The upper electrode of the cell features a profiled structure in the form of two high-aspect-ratio nanoscale
tip structures (HANTS), where one tip has a radius of 10 nm, and the radius of the second tip varies in the
range of 10 to 50 nm. Platinum was chosen as the material for the upper electrode due to its unique physicochemical
properties, including high chemical inertness across a wide range of temperatures and aggressive
environments, low electrical resistivity, and resistance to oxidation. These characteristics make
platinum an optimal material for use in electronic devices and sensor systems where long-term stability
and minimal energy losses during signal transmission are required. The results of modeling the electric
field strength distribution in the interelectrode gap of the memristive cell are presented. The modeling was performed using COMSOL Multiphysics software, which solves systems of nonlinear partial differential
equations using the finite element method, with a potential difference of 5 V between the electrodes. Based
on the modeling results, the dependencies of the electric field strength on the geometric parameters of the
memristive cell were obtained and analyzed. Local enhancement of the electric field strength was identified
along the perimeter of the oxide-HANTS interface. The increase in the non-uniformity of the electric
field strength grows with the thickness of the oxide film and can reach 13.4%. The obtained results can be
used in the development of neuromorphic electronic components for robotic systems and artificial intelligence
systems based on memristors

In the context of the need to ensure food security, the tasks of optimizing production processes in the
agricultural sector are becoming relevant. For example, given the shortage of labor in agriculture, it is
necessary to develop and implement robotic systems to automate the processes of plant care, harvesting
and processing. The article presents the results of the development of an autonomous robot for picking
apples, created on the basis of a universal anthropomorphic robot developed at the Kabardino-BalkarianScientific Center of the Russian Academy of Sciences. The robot is equipped with two multi-link manipulators
similar to human hands, which allows it to perform complex harvesting tasks. To ensure intelligent
control of the entire system, a multi-agent neurocognitive architecture is used, which imitates the work of
the human brain and allows the robot to adapt to changing environmental conditions. The robot is
equipped with a set of sensors, including video cameras, ultrasonic and infrared rangefinders, lidar and
encoders on the manipulator drives. This allows it to accurately determine the location of apples, assess
their ripeness and plan the trajectory of the manipulators. Particular attention is paid to the development
of a gripper that imitates a human hand and allows you to adjust the squeezing force, which minimizes the
risk of damage to the fruit. A multi-agent neurocognitive architecture is used to control the robot, which
provides autonomous decision-making based on sensor data. The system is able to build a map of the area,
determine the position of the robot and plan a route, as well as recognize apples and assess their condition.
The article also considers the problems associated with the automation of harvesting in agriculture,
including a lack of labor and crop losses due to improper operation of equipment. The authors emphasize
that automation and robotization of harvesting processes have great potential, especially for crops
that require an individual approach, such as fruits and vegetables. The presented robot demonstrates high
efficiency in solving these problems, which is confirmed by the results of field tests. The developed system
can be adapted to work with other crops, which makes it a universal solution for the agricultural industry

The paper presents the results of a study of the possibility of forming and executing a mobile robot
operation scenario using neuromorphic information and control elements (logical elements "AND", "OR"
and "NOT"; neuromorphic extrapolator; neuromorphic emergency detector; neuromorphic conditioned
reflex formation mechanism). A brief description of these information and control elements is given. In
this case, the segmented spike model of the CSNM neuron with the possibility of structural learning acts as a basic element. The trained CSNM model is capable of solving the binary classification problem, which
implies the possibility of using it as a separate information control element – a state detector. It is proposed
to organize the execution of the mobile robot operation scenario on the basis of universal
neuromorphic modules using the specified neuromorphic information and control elements. The proposed
task management mechanism boils down to the following. Universal neuromorphic modules used as switch
blocks are prioritized. The detection of a particular situation is performed by means of a universal
neuromorphic module, the priority of which is higher, which leads to the inhibition of all universal
neuromorphic modules, the priority of which is lower than this one. By adding or removing universal
neuromorphic modules, as well as changing their priority, we get the desired behavior of a mobile robot.
The paper presents the results of a study of the proposed mechanism for organizing a robot operation
scenario on both a computer model and a real robot. A special emulator was developed for computer
modeling, which made it possible to comprehensively investigate the proposed solution. Moving towards
the target in a room with partitions was chosen as a test task. The experimental results have shown the
fundamental applicability of the proposed mechanism for forming and executing a robot operation scenario.
The paper highlights the main disadvantages of the current implementation: the possibility of situations
leading to looping of a set of actions performed by a mobile robot, as well as the need for manual coordination
of the main parameters of the scene and the mobile robot (metric characteristics of the specified
size of the scene zones, angular and linear velocity of the mobile robot, etc.). The ways to eliminate these
disadvantages are proposed. The conclusion is made about the prospects of using the segmented spike
model of a neuron and networks based on it for the tasks of controlling mobile robots

The present study is aimed at systematizing scientific knowledge about diseases of agricultural
crops with the subsequent integration of the data obtained into automated agricultural production management
systems. The relevance of the work is due to the need to minimize economic losses in crop production
through early diagnosis of pathologies and optimization of phytosanitary control. As part of the study, a classification of plant diseases was carried out.The basil plant (Ocimum basilicum L.), characterized
by high susceptibility to phytopathogens under intensive cultivation conditions, was chosen as a model
object. To create an automated diagnostic tool, a specialized dataset was collected, including 214 images
of basil at various stages of vegetation. The shooting was carried out under controlled conditions
using an RGB camera. Each sample is annotated with the localization of damage and the affected area.
Special attention is paid to the methodological aspects of the formation of data banks for biological systems.
It has been established that the key problems are the high variability of morphological features in
plants, the influence of environmental factors on the visual manifestations of diseases. Based on the analysis
of the data obtained, the architecture of the early warning system is proposed, which includes three
modules: a sensor unit – small cameras and microclimate sensors. The algorithmic block is a neural network
model for semantic image segmentation and algorithms for assessing the dynamics of pathology
development. The decision – making and notification interface provides recommendations for adjusting
irrigation regimes, applying pesticides and trace elements. The convolutional neural network is trained
based on the YOLOv11 framework using data augmentation methods (Gaussian noise, affine transformations)
and transfer learning. Validation of the model on the test sample showed a detection accuracy of
74.7% (F1-score = 0.72). To reduce false positives, postprocessing of predictions has been implemented,
taking into account the spatial and temporal correlation of the data. The developed prototype demonstrates
the potential of integrating computer vision and agronomy to create predictive control systems.
Further research is planned to expand the dataset and increase parametrs, as well as the introduction of
data processing algorithms on edge devices to reduce delays in decision-making. The results obtained can
be adapted for other indoor crops, which contributes to the development of precision agriculture and reduces
anthropogenic stress on agroecosystems

The paper describes a hybrid algorithm for situational trajectory planning under partial uncertainty for a
two-dimensional space based on the integration of the wave and ant algorithms, which allows constructing trajectories
of minimum length in real time with simultaneous optimization of a number of other quality criteria for
the constructed path. The processes of forming a trajectory section and moving an object along it alternate at
each step. The trajectory is formed sequentially (step by step) at two levels of each step. The local visibility zone
and the region covered by it on the terrain map are formed and oriented relative to the current reference vector.
The first-level procedures sequentially form a chain of pairwise adjacent regions with localized obstacles on the
terrain map in steps. The second-level procedures form a set of trajectories for the passage of a moving object
through a region at a step. When the chain of regions merges, a terrain region is formed through which the trajectory
is laid. The entire trajectory is a set of individual trajectories for the passage of a moving object through
regions connecting its initial position with the target position. The search for a solution is carried out by a population
of agents on a solution search graph. The vertices of the set correspond to the cells of the region. Two
vertices are connected by an edge if the corresponding cells on the terrain model in the form of a discrete working
field are adjacent and the transition of the connection from one cell to another is possible. It should be noted
that the synthesis of the trajectory and the movement of a moving object under uncertainty is a complex task that
requires the integration of various sensor systems, data processing algorithms, path planning algorithms and
motion control systems. The constant development of technologies in the fields of artificial intelligence, machine
vision and robotics allows the creation of increasingly sophisticated autonomous navigation systems. However,
complete autonomy and guaranteed safety of a moving object under any conditions still remain complex tasks
for research.

The article presents the process of developing an underwater vehicle simulator (USV) with an installed
5-degree underwater manipulator complex (MC). The simulator is designed for complex testing of
autonomous interaction of a marine robotic complex (MRC) with underwater infrastructure objects. In
particular, an example of solving the problems of simulator operation with a model of an underwater panel
of an underwater production complex (UPC) and solving the problem of determining concretions and
their autonomous collection using the simulator and MC are considered. Modern trends in the development
of underwater robotics are focused on the creation of resident autonomous systems capable of operating
in remote and hard-to-reach areas of the World Ocean all year round. The development of resident
technologies is associated with the need to reduce operating costs, minimize risks to personnel and increase
the autonomous functioning time of underwater complexes. The use of such technologies is especially
relevant in the conditions of offshore shelf development, where traditional methods of operating
underwater vehicles encounter technical and economic limitations. The need to carry out work on the distant shelf is due to the increasing demand for hydrocarbon resources and the depletion of easily accessible
deposits on the continental shelf. According to forecasts, promising deep-water areas located at
depths greater than 1000 m have significant potential for oil and gas production. According to experts, the
volume of recoverable reserves in such areas can amount to hundreds of billions of barrels of hydrocarbon
raw materials, which makes the development of effective autonomous solutions a strategically important
task for the oil and gas industry. The paper presents software and hardware solutions used in the
implementation of the USV. A structural diagram of the design is provided; the software architecture and
features of the use of artificial neural network (ANN) systems as part of the technical vision system (TVS)
of the USV are described. The use of TVS allows to significantly increasing the autonomy of underwater
manipulators when performing complex technological operations, such as capturing objects from the
ground, working with bottom infrastructure objects, etc. In conclusion, the obtained results are demonstrated,
confirming the operability of the adopted design, software and hardware solutions when performing
real work in autonomous mode with mock-ups of hot-stab and torque-tool working tools and mating
parts located on the mock-up of the UPC panel.

The work is devoted to solving urgent problems of joint autonomous visual navigation for air and
ground-based robots in urbanized environments. These environments are highly demanded for special
operations, including dense urban areas and buildings, where the use of traditional remote control devices
is limited due to the presence of shielded areas. The proposed solution addresses group navigation tasks
based on data from onboard vision systems during operational reconnaissance of the working area by an
unmanned aerial vehicle (UAV). The results of this reconnaissance enable autonomous movement and
flight, both for individual heterogeneous robotic systems and for groups.The navigation algorithms are
based on methods for extracting a horizontal reference surface and horizontal sections of the external
environment from a volumetric point cloud generated by an onboard lidar. These methods allow for the
precise and rapid determination of all six coordinates of the control object. Cases where the navigation
task cannot be fully solved due to specific environmental characteristics are also considered. To address these challenges, methods are proposed to enhance lidar rangefinding data by integrating video camera
data. An accuracy assessment of the video navigation solutions is provided, obtained through mathematical
modeling of the external environment and the generation of video data. To ensure safe autonomous
flight and movement of robotic systems in urban environments, methods for reducing video navigation
errors are proposed. These methods utilize a specially designed bank of reference images with known
coordinates of their formation. The effectiveness of the applied methods and the proposed video navigation
algorithms is confirmed by experimental studies of the corresponding software and hardware in real
urbanized environments

With the growing number of cars and limited space, many cities are realizing the importance of implementing
intelligent parking systems to improve urban mobility and convenience for drivers. The level of
implementation of intelligent parking based on various technological solutions is growing, but to achieve
maximum efficiency, it is necessary to continue to develop technologies, integrate them with other systems
and take into account the needs of users. The purpose of the study is to develop a multi-agent intelligent
system for monitoring and managing parking space reservations in the city parking network. The architecture
of a multi-agent intelligent parking management system has been developed, which provides automatic
access control to parking spaces taking into account the wishes of parking lot owners, driver orders, the traffic
situation in the city and safety requirements. The main element of the developed system is parking, which
is represented by a set of parking spaces equipped with automated parking space management systems (parking
attendants), a communication system and data collection tools (surveillance camera and weather stations).
Parking spaces and parking attendants are managed by an intelligent control system based on multiagent
neurocognitive architectures. A prototype of a hardware and software complex of a multi-agent intelligent
parking space management system has been developed in the form of a client-server architecture.
The server is responsible for collecting, processing, storing data and managing automated parking attendants.
Two types of clients are connected to the server - a mobile application of the administrator and the
driver. The administrator has the ability to manage parking (set fixed prices or use server recommendations,
book parking spaces for employees) and view statistics (current load, parking statistics, data on accepted
payments, parking work forecast, recommendations). The driver has the ability to view the status of parking
in the area of interest (number of free spaces, waiting time for a free space, cost, recommendations for the
most convenient parking) and book a parking space with the ability to pay online

Currently, the field of underwater robotics is actively developing to solve applied and research
problems. One of the promising areas of underwater robots’ application is the implementation of
bioinspired type of swimming. The use of autonomous bioinspired underwater vehicles (BUV) will potentially
expand the scope of application of low-noise and safe for local fauna underwater robots for monitoring
and exploring the terrain. The aim of the work is to develop and test a methodology for model-based
design of a motion control system for biomorphic underwater robots. In this work a typical BUV design
with oscillatory type of swimming is considered. Problematic issues of modeling the BUV dynamics, as
well as the synthesis of their control systems are described. For BUVs with oscillatory types of swimming,
typical technological operations are identified. Typical technological operations are chosen based on the
design features of the BUVs and the composition of their propulsion and steering complex. A control system
design methodology based on the combined use of numerical modeling technologies and classical
automatic control theory is proposed. Based on the proposed methodology, numerical hydrodynamic
BUV’s models with oscillatory types of swimming are developed. Identification computational experiments
are conducted. The transient processes which characterize BUV’s dynamics during the performance of
each typical operation are defined. Based on the simulation results, cybernetic simplified models of BUV’s
based on the typical blocks of the automatic control theory are developed. Based on the cybernetic models,
based on the numerical optimization a synthesis of BUV’s control system in accordance with the proposed
methodology is performed. The developed algorithms are tested based on numerical hydrodynamic simulation
results. Possible prospects for the use of the BUV’s are formulated.

Trends in the development of modern robotics and the needs of practice require an increase in the
degree of autonomy of robotic complexes. Increasing the degree of autonomy, in turn, requires an increase
in situational awareness and, as a result, an increase in the volume of data and the efficiency of
their processing in real time on on-board resources. At the same time, the requirement of economic feasibility
of the proposed solutions remains. Taking into account the fact that in domestic practice, in most
cases, remote-controlled robots are used, it becomes necessary to increase their autonomy using existing
technical solutions. This direction of development of robotic complexes is called the transition from remote
control to supervisory control. Along this path, an increasing number of information management
functions are transferred from the operator to the on-board information management system. Based on the
analysis of the world and our own experience in the development of robotic complexes, the author's methodology
for creating robots with an increased degree of autonomy, we have identified a key element in
ensuring the intellectual autonomy of mobile devices. A unified software and hardware module for information
support of mobile robots is proposed. The module is based on a vision system with an open software
and hardware architecture. This module allows increasing the degree of autonomy of ground-based
robotic complexes in terms of intellectual autonomy gradually, while remaining within the framework of
economic feasibility. The open software architecture of the module takes into account the de facto variety
of hardware solutions in existing remote-controlled mobile vehicles and allows for an increase in the degree
of autonomy - to switch from remote control mode to supervisory control step by step, according to
the tasks being solved and the available means. A technique for creating new or reengineering existing
RTK samples is proposed. The methodology includes an analysis of the overall layout of the RTK with an
emphasis on the software and algorithmic part of the on-board information and control system. This takes
into account the conditions for matching the requirements for the sensor and computing parts. The paper
considers examples of the application of this technique to the improvement of existing samples of groundbased
RTCs.. Practical tasks and examples of their solution using the proposed module are presented

The aim of the research is to develop a methodology for identifying and determining the location of objects
under conditions of low visibility and potential changes in their shape, with a focus on extracting parts
created using selective laser sintering (SLS) from a powder medium. The study examines two fundamentally
different approaches to forming control algorithms for a robotic manipulator. The first approach, trust-based, is
based on the assumption of minimal displacement of the object during manipulation. The manipulator moves
along a trajectory calculated from a preliminary three-dimensional model without correction until the moment
of capture. This method is characterized by high operational speed and minimal computational costs. However,
it carries risks such as object deformation due to environmental resistance, displacement of the part upon contact
with the tool, and the inability to capture the object if it deviates significantly from its nominal position.
The second approach, cautious, involves the gradual removal of powder layers to visualize the object and adjust
the trajectory before capture. This method includes several stages: removing the top layer of the medium to
partially expose the part, analyzing data to refine the object's position, and constructing an adaptive trajectory
considering possible displacement. Special attention in the article is given to data generation for training neural
networks, which are used for object identification under noisy conditions. Two methods of artificial modeling of
powder coatings are considered. The primitive method involves expanding the vertices of a three-dimensional
model along their normals with the addition of random noise. The improved method proposes differentiated
powder distribution considering local surface curvature. Subsequent experimental results showed that training a
neural network using real data has low efficiency. Recognition accuracy ranged from 60% to 75%, which is
attributed to the small sample size and the influence of external factors such as lighting and interference. At the
same time, the use of synthetic data, prepared according to the methodology presented in the study, increased
recognition accuracy to 92%. The practical significance of the work lies in the development of a methodology
for searching, detecting, and identifying a part immersed in powder, which can be used to automate postprocessing
processes in industries utilizing selective laser sintering. The developed solutions are adapted for
integration into robotic systems operating under conditions of limited visibility. The proposed methods can be
scaled to a wide range of tasks in additive manufacturing and robotics, making them promising for implementation
in industrial processes.

The purpose of the study is to improve the quality of operational planning (program control) of logistics
processes in the conditions of modern urban systems with the interaction of a group of robotic systems.
The quality of management in this study will be assessed by the number of deliveries completed after
established directive deadlines. The goal set during the study is decomposed into the following tasks: system
analysis of the current state of research in the field of metropolitan logistics, implementation of a
substantive and formal formulation of the problem of operational planning of logistics processes in a metropolis
using a group of robotic complexes, development of a model and algorithm for operational planning
of logistics processes in a metropolis using a grouping of robotic complexes, development of special
model-algorithmic support and its software prototype for solving the problem of operational planning of
logistics processes in a metropolis using a grouping of robotic complexes. Proactive (anticipatory) management
of a group of robotic systems when solving transport and logistics problems in a metropolis within the framework of the “Smart City” concept allows increasing the economic efficiency of cargo delivery.
The article examines the scientific and technical problem of synthesizing technologies (plans) for the timely
delivery of small-sized cargo using a group of robotic systems. The scientific significance lies in the
application of the concept of integrated (system) modeling and proactive (anticipatory) management, and
the practical significance lies in ensuring timely delivery of goods using a group of robotic complexes in a
metropolis. The article discusses an example of solving the problem of operational planning of logistics
processes using the example of Innopolis using the characteristics of Yandex delivery robots (as robotic
complexes). During the study, an analysis of various options for objective functions was carried out: maximizing
profit and minimizing delivery time; profit maximization; minimizing time; minimizing the number
of robotic systems. The following indicators were chosen to evaluate the results obtained: total profit from
deliveries; the number of deliveries not delivered on time and the total number of completed orders.
The most suitable objective functions for solving the problem are time minimization or simultaneous time
minimization and profit maximization. In addition, the conclusion provides directions for further research

An algorithm for adaptive control of a DC motor based on the use of machine learning technology
with reinforcement is proposed and investigated. An overview and brief analysis of the state of affairs in
the field of intelligent motor control systems is given. A mathematical model of the DC motor is presented,
and a structural scheme for training an intellectual agent is presented. An intelligent adaptive motor speed
control system is proposed. The DC motor is represented as a black box with the limited input and output.
The control system is based on a zero-order Q-learning algorithm. It is assumed that the output of the
intelligent agent is a control applied to the motor input. The intelligent system uses a tabular approximation
of the value of each of the control action. In this article, we study the effect of the discreteness of the
representation of state, the set of control effects used, the applied rewards, and the parameters of the
learning algorithm on the control error. The sensitivity of the control system to the parameters of the motor
and an unmeasured moment is investigated. Based on the results of the study, a modified algorithm is
proposed, which assumes the measurement or evaluation of the current of the motor stator. The control
algorithm provides robustness to parameters and external disturbance. Additionally, the approximation of
the control value function using polynomials and using a neural network are investigated

The article discusses a trajectory planning method for a group of mobile robots that ensures safe
movement and eliminates the possibility of collisions both between the robots themselves and with external
obstacles, including moving objects. The developed mathematical model considers three main collision
scenarios: intersection of robot trajectories within the group, interaction with stationary obstacles, and the probability of collision with moving objects. Each of these scenarios is analyzed in detail to ensure
maximum safety during movement, and their consideration allows for efficient adaptation of robot routes
to changing environmental conditions. The trajectory of each robot is represented as a piecewise linear
path with intermediate points, which are optimized to ensure safe movement. Special attention is paid to
speed adaptation on different segments of the trajectory: a robot can adjust its speed based on current
conditions to minimize the risk of collisions. To evaluate distances between objects, the Euclidean norm is
used, allowing for the calculation of minimum distances between the centers of spherical representations
of robots and obstacles. The problem is solved in two stages. In the first stage, a trajectory is constructed
for the first robot, taking into account initial conditions and obstacle placement. In the second stage, trajectories
are formed for the remaining robots, considering the already planned routes. For optimizing the
coordinates of intermediate points and speeds, a genetic algorithm is applied, which minimizes travel time
while ensuring safe movement. The genetic algorithm uses crossover and mutation operators to generate
diverse solutions and performs checks to ensure compliance with safety conditions. Numerical simulations
were conducted using Python, with the Matplotlib library used for visualization of results. During the
experiments, 50 tests were performed with varying numbers of obstacles (from 5 to 10). Analysis of the
results showed that as the number of obstacles increased, both the computation time and the quality of the
generated trajectories improved. This confirms the effectiveness of the proposed method for controlling
groups of mobile robots in dynamically changing environments

One of the key parameters of any modern mechanical system is its vibration and acoustic characteristics,
which have a direct impact on the environment and humans during operation. In this connection, the
task of diagnosing the vibration characteristics of various complex mechanical objects, to which industrial
robotic complexes can be referred, remains relevant. Due to the difficulty in carrying out diagnostics and
experimental debugging of newly developed mechanisms, it is interesting to apply modern approaches to
solving the problem of diagnostics, in particular, with the use of neural networks and numerical methods.
The purpose of this work was to investigate the possibility of joint application of spike neural network and
finite element method for estimation of vibration characteristics on the example of wave gearbox bearing.
The paper describes in detail the algorithm of diagnostics, which includes the stages of development of both
the finite element model of the investigated mechanical system and the development of the neural network
architecture. At the same time, the generation of training and control datasets for the neural network is carried
out on a simplified finite element model having characteristics similar to the detailed one, which is ensured
by the coincidence of the first ten eigenforms of the assembly. The data sets were generated on the
basis of numerical calculations using an explicit scheme of integration in time of a simplified model of a
gearbox with several types of artificially introduced defects similar to those appearing during operation of a
real bearing. To analyze the frequency characteristics, a spike neural network architecture was developed
and further improved on a training set of single defects. As a result of the study it was determined that the
developed spike neural network provides classification of data on the control dataset with 85% accuracy,
which allows us to conclude about the applicability of the proposed method of determining the vibration state
of mechanical systems with the joint use of neural networks and finite element method.

In this research work, the problem of task allocation in a multi-agent system is considered, where
each agent is a robot, and each task is represented by a position, which should be visited by one agent.
This problem is very similar to the multi-agent traveling salesman problem, which, unlike the famous traveling
salesman problem, involves several traveling salesmen who visit a given number of cities exactly
once and return to the starting position with minimal travel costs. Therefore, the multi-agent traveling
salesman problem is analyzed as a representative of the task allocation problem. The multi-traveling
salesman problem is important for the field of route optimization and task allocation between several
agents. It includes two different, but interrelated subproblems: distribute cities among agents and determine
the order in which each agent visits cities. In the literature, there are 3 concepts for solving this
problem with respect to solving its two constituent subproblems: the optimization concept, where both
subproblems are solved simultaneously; The Cluster-First, Route-Second concept is where the question of
which tasks to assign to which salesman is first decided, and then the question of the order in which each
salesman solves his tasks is decided; The Route-First, Cluster-Second concept is where the question of the
order in which tasks should be visited is first decided, and then this cycle is divided between agents without
changing the order of visits in order to answer the question of which tasks each agent takes on. This
paper proposes a hybrid approach to solving the multiple traveling salesman problem (mTSP), which
combines the ideas of two well-known concepts: "First clustering, then routing" and "First routing, then
clustering" in order to obtain their positive aspects and get rid of their weaknesses. To evaluate the effectiveness
of the developed method, a comparative study was conducted using the classical method for solving
the multi-traveling salesman problem. The results were evaluated based on three key criteria: the
computational time to obtain a solution to the multi-travelling salesman problem, the total length of the
routes travelled by the salesmen, and the maximum route length among them. The analysis of the experimental
data showed that when using the proposed method, the maximum path length among the routes
travelled by the agents (load imbalance) is reduced by an average of 26%.

The study is devoted to improving the subsystem of information protection in the radio channels of a
cyberphysical system using the example of a robotic complex (RTC). Modern and promising critical conditions
for the use of RTCs are considered, which determine the sets of requirements for the characteristics
of both RTCs and their subsystems, such as the radio data transmission system (RS) and the information
security subsystem. One of the approaches to meeting the requirements is the unification of theseRTC subsystems, which can be divided conditionally into two scientific and technical tasks: unification of
radio protocols and unification of information security tools in RS radio channels. The paper presents the
practical problems obtained as a result of the analysis, which lie at the intersection of two areas of research
– RS and information security subsystems. A hypothesis has been formed about the potential for
effective resolution of one of these practical problems – providing an information protection system with
cryptographic keys - by including a cryptographic key generation subsystem (CKGS) from biometric data
used as the initial key information in the RTC information protection system. The proposed improvement
has several aspects – regulatory, economic, and technical. The paper examines only the scientific and
technical side of the issue, as a result of which a functional model of the CKGS is proposed, which provides
a study of the possibilities of the modeled subsystem for the implementation of the formulated principles
of functioning. The purpose of the work is to develop a model of the CKGS functioning for the cryptographic
information protection system in the RS RTC radio channels and the formation of its algorithmic
content. The object of research is a system of cryptographic information protection in RS radio channels.
The subject of the research is an algorithm for generating cryptographic keys for a cryptographic information
protection system in RS RTC radio channels. To achieve this goal, a class of abstractions involved
and a methodological apparatus are substantiated that uses the provisions of the theory of algorithms to
prove the existence of an algorithm that solves a formulated mass problem and has specified non-trivial
semantic properties. Research methods – analysis, analogy, synthesis, decomposition, abstraction. The
main mass problem and the hypothesis of its solvability are formulated. In order to test the hypothesis, the
corresponding theorem is formulated and proved. The proposed model makes it possible to prove the joint
effective feasibility of various information processing algorithms

The task of detecting obstacles in front of a mobile robot has been successfully solved long ago using
laser and ultrasonic sensors. However, obstacles that are not detected by these types of sensors may endanger
the safety of the robot. To detect them in the work, it is proposed to use a technical vision system
(STZ), the information from which is processed by a semantic segmentation neural network, which returns
the mask of the obstacle on the frame and its class. The basis for such a network was the SAM universal segmentation
network, which requires further development to be applied to the semantic segmentation task.
The peculiarity of this network is its universal applicability, that is, the ability to select any objects in any
filming situation. At the same time, SAM does not predict the semantics of the object. In this paper, an additional
module is proposed that makes it possible to implement semantic segmentation by classifying the features
of the selected objects. The possibility of using such a module to solve the problem of supplementing the
network output with new information is substantiated. The classification result is then fed into the same filtering
algorithm as the masks to ensure consistency between the result of the universal network and the complementary
module. After integrating the module with the model, a new semantic segmentation model was
obtained, called RTC-SAM in the work. It was used to perform semantic segmentation of a publicly available
dataset with images of an open area. The 45% result obtained by the IoU metric exceeds the result of existing
methods by 13%. The images of the results of using the new network shown in the work make it possible to
verify its performance. It also describes the testing of the developed solution with a study of the performance
of the developed model on a PC and a mobile computer. The algorithm on the mobile computer shows insufficient
speed to enter real-time mode – more than 3.5 seconds to process one frame. In this regard, one of
the directions of further research in the field of improving system performance.

The purpose of this study is to develop an improved model of a magnetically gradient measuring
system, which plays a key role in solving the problem of simultaneous estimation of the parameters of the
Earth's magnetic field (carrier) and the parameters of the anomalous magnetic field. The relevance of the
study is due to the need to optimize the process of differential magnetometry. Differential magnetometry
allows you to extract useful information from a magnetic field without the need for time-consuming and
lengthy detailed mapping of the area. This is achieved by estimating the parameters of the magnetic field
gradient, which significantly increases the information content and efficiency of determining the location
of sources of magnetic anomalies. However, measurements of physical fields are significantly affected by
various disturbances that may occur due to equipment and design errors, as well as due to their magnetic
properties. These interferences distort the measurement results and reduce the accuracy of determining
the magnetic field parameters. In this regard, the development of a model that takes into account the effects
of interference and compensates for their effects is an extremely important task. The proposed model
of a magnetically gradient measuring system will take into account the effects of interference and compensate
for their effects, which will significantly improve the accuracy of estimating magnetic field parameters.
In addition, the model will help solve the problem of jointly estimating the parameters of the carrier
field and the constants included in the measurement model. This will increase the efficiency of differential
magnetometry and make it more applicable in various fields. In particular, the improved magnetically
gradient measuring system will be useful in geophysical mapping, mineral prospecting, environmental monitoring, as well as in specific combat conditions. For example, in geophysical mapping, it will allow
you to more accurately determine the boundaries of various geological structures, which is important for
the search for minerals. In the conditions of clearing the liberated territories of hidden objects left by the
enemy, such as mines or underground structures, the system will help to identify them quickly and accurately,
which will significantly increase the safety of military personnel and civilians. Thus, the development
of an improved model of a magnetically gradient measuring system has a wide range of potential
applications and is an important step in the development of differential magnetometry technologies

Satellite navigation equipment often operates under conditions of a priori uncertainty of the parameters
of the mutual dynamics between the transmitter and the consumer and the signal-to-noise ratio of the
received signals of satellite radio navigation systems. Classical Bayesian algorithms for Phase lock loop
system require a priori knowledge about the parameters of the phase process dynamics and the signal-tonoise
ratio (SNR) of the received signals. As a result, the operation of such algorithms under conditions
other than that a priori specified is not optimal to the criterion of minimum error variance. Moreover, a
sudden change in the signal-to-noise ratio or dynamics can lead to a tracking failure in such a system.
The purpose of this work is to develop an optimal phase tracking system that is adaptive to the dynamics
of the phase process and the signal-to-noise ratio in order to maintain phase tracking in the widest possible
range of operating conditions while tracking global navigation satellite system signal. An adaptive
multichannel phase lock loop system has been synthesized as a result of formulation and solution of signal
processing problem in terms of the statistical synthesis theory. Adaptivity to the changing power of the
received signal is achieved by including the signal-to-noise ratio [dBHz] in the vector of estimated filter
parameters. Adaptability to the intensity of the phase change dynamics is achieved through the use of a
multi-channel filtration system. Statistical modeling of an adaptive multichannel phase lock loop tracking
system with a complex algorithm for tracking the code delay of the signal of satellite radio navigation
systems has been carried out. The values of the sensitivity of phase tracking under various dynamic conditions
are determined. The adaptive multichannel phase lock loop system is able to withstand an signal-tonoise
ratio jump from 50 to 10 dBHz and back without loss of phase tracking in low dynamics conditions
(only the drift of the reference quartz oscillator). The AMPLL system is able to withstand abrupt transitions
of dynamics between low and high (the sinusoidal acceleration 10g and sinusoidal jerk 10 g/s) without
loss of phase tracking under the 24 dBHz signal-to-noise ratio. Thus, in real conditions, when the dynamics
of the GNSS receiver and the SNRs of the received signals change in an unpredictable way, the
AMPLL system keep tracking in a much wider range of conditions than the non-adaptive PLL

The objective of this work is to create a system for detecting underwater objects intended for installation
on surface platforms (aircraft or remotely piloted aircraft). Systems of this type can be used to solve
a wide range of problems in various areas of the national economy: searching for rare fish and marine
mammals, determining their migration routes, diagnostics and laying underwater pipelines and fiber optic
cable networks, monitoring seawater pollution, searching for sunken ships and archaeological treasures,
and carrying out rescue operations. To solve this problem, the process of laser radiation propagation to
an object across the air-water interface was described, a rough sea surface was modeled, and a number of
mathematical assumptions and approximations were proposed. In the practical part, a structural diagram
of a laser optical-television active-pulse underwater vision system was developed, including receiving and
transmitting channels, as well as a control device consisting of an image processing unit and a control controller. The receiving channel includes an electron-optical converter of the III+ generation, highly
sensitive in the spectral range of sea water transparency. The main element of the transmitting channel is
a highly efficient pulse laser emitting in the spectral range of sea water transparency. The assembled device
has undergone field tests, as a result of which it became clear that detection and recognition of underwater
targets from an aircraft through the air-water interface using the generated image is possible,
the maximum detection range and recognition of underwater targets of the active-pulse underwater vision
system from an aircraft through the air-water interface is mainly determined by: attenuation of optical
radiation in sea water and the power of the illuminating laser pulse radiation, At the same time, a distinctive
feature of the active-pulse underwater vision system is that the increase in the detection and recognition
range is almost directly proportional to a certain level of laser radiation power, and a further increase
in power leads to an insignificant increase in range

The article addresses the problem of visual navigation for unmanned aerial vehicles (UAVs), which
involves the automatic determination of the current position of the UAV (coordinates in the ground (local)
coordinate system) based on the comparison and identification of descriptions of the current images (CI)
received on board with reference descriptions stored in the form of a digital map in the memory of the
UAV's onboard computer. The aim of this work is to improve the efficiency of visual navigation methods in
terms of increasing computational performance, robustness, and accuracy of image identification algorithms
in complex and changing observation conditions by using semantic descriptions of observed scenes.
In this work, semantic descriptions are understood as descriptions that include classes of objects observed
in the scene, their attributes, and relationships between them. The preparation of semantic descriptions of
the map is carried out at the pre-flight preparation stage of the UAV using pre-trained neural networks for
semantic segmentation. Semantic descriptions of the received CIs are generated on board the UAV. The
use of neural network algorithms allows this process to be implemented in real-time for a wide range of
observation conditions (different times of day and year). The use of semantic descriptions of the map and
CI reduces computations compared to traditional pixel-by-pixel matching of raster images. Semantic descriptions
are compared by matching object classes, their attributes, and relationships. The work presents
a general algorithm for visual navigation, the main stages of the methodology for forming semantic descriptions,
and the algorithm for comparing and identifying semantic descriptions of CIs and map descriptions.
A hierarchical algorithm for comparing and identifying images based on the sequential application
of semantic and raster descriptions of observed scenes is proposed. It is shown that the use of the procedure
for comparing semantic descriptions of CIs and maps by the classes of objects present significantly
reduces the computations necessary for image identification

The aim of the study is to provide the possibility of UAV navigation when it is impossible to use satellite
global positioning systems using GNSS in electronic warfare conditions. To achieve this goal, a
comprehensive approach to ensuring UAV navigation by visual landmarks using machine vision systems is
proposed. It is proposed to synthesize images of the underlying surface by combining sensor data, which
improves the quality of UAV positioning in the absence of satellite navigation systems. It is shown that
when combining remote sensing images of different origin and changing external operating conditions
(day-night, winter-summer, etc.), it is important to most fully localize the objects of the underlying surface.
In the machine vision system for visual navigation of UAVs by natural landmarks, a method of electromechanical
image scanning is proposed, which allows increasing the field of view of a camera of an arbitrary
range. Modeling of the characteristics of the machine vision system with electromechanical scanning
is carried out to determine the limits of applicability to the problem of visual navigation. It is shown that
the most significant parameter of positioning accuracy is the shooting height of the underlying surface,
which is quasi-linear under the condition of a fixed camera tilt angle, and for high-quality positioning, the
best option is the frontal position of the camera at the nadir point. The proposed approach allows creating
virtual 3D models of the underlying surface, thereby increasing the capabilities for more accurate recognition
of objects based on the scale and size of the segmented areas. Measuring the camera elevation angle
can be used to detect and recognize natural landmarks that can be predetermined (road intersections buildings or structures, utility facilities, etc.). On the other hand, the frontal position of the camera with a
zero elevation angle is advantageous for verifying the flight route, positioning the UAV relative to the
reference landmark. This is due to the fact that with the widespread use of software based on mathematical
models, the photogrammetric ruler technology has become appropriate for quantitative measurement
of terrain plans and maps

In the development of modern robotic complexes (RC), there is a significant diversity in both hardware
and software solutions, which creates additional challenges in selecting a rational hardware and
software composition to ensure the required computational power and to effectively address the assigned
tasks. On one hand, it is often necessary to work with an already installed set of computing systems (CS)
that form the onboard computing system (OCS) of the RC, which substantially limits the possibilities for
modifying the software composition and necessitates the adaptation of algorithms to fixed hardware resources.
On the other hand, when there is an opportunity to modify or create a new hardware composition,
it becomes necessary to choose a hardware configuration that can meet the computational requirements
of the tasks being solved. This article proposes a methodology for assessing the hardware composition
of the OCS of RCs based on the tasks being solved, relying on the use of multiversion programming
and the creation of solution passports. Each variant of the software solution for a specific task is supplemented
by a structured passport that contains both quantitative and qualitative characteristics, allowing
for a detailed comparative analysis. Based on these solution passports, a mathematical model is developed
that enables the selection of a set of computing devices capable of executing all the assigned tasks while
simultaneously minimizing the total cost, energy consumption, or other operational characteristics of the OCS. Mathematically the problem under consideration is reduced to a generalized multiplicative multidimensional
knapsack problem with multi-choice and additional constraints, which allows both resource and
topological dependencies among the tasks being solved to be taken into account. Experimental results obtained
using the developed simulation platform are presented, demonstrating the practical applicability of the
methodology and confirming the possibility of using it to obtain quantitative estimates of the hardware composition
variants of the OCS of RCs. This approach can be adapted for various types of RCs, which facilitates
its use in related studies in the field of optimizing computing systems for robotic complexes

The article examines the possibility of using convolutional neural networks for technical object
recognition in the context of radio monitoring. The focus is on the development and optimization of algorithms
for processing radar signals using deep neural networks. Studies have shown that the use of CNN
can significantly improve the classification accuracy of radio signals compared to traditional processingmethods. The developed approach is based on the extraction of hierarchical features from spectral images
of radio signals and their subsequent classification using a trained neural network. The paper presents the
results of experimental studies conducted on a dataset of more than 10,000 samples of radio signals of
various types. It is shown that the proposed technique ensures recognition accuracy of up to 94% when
working with noisy signals and the probability of a false alarm is no more than 0.05. Special attention is
paid to the choice of neural network architecture for the specifics of the radio monitoring task. The options
for converting radio signals into a spectral image for real-time processing were also considered in
detail. Data preprocessing methods have been developed, including amplitude normalization, frequency
correction, and interference elimination. The results of the study can be used in radio broadcast control
systems and to ensure electromagnetic compatibility of electronic devices. The results obtained demonstrate
the prospects of using CNN in the tasks of technical recognition of radio monitoring objects and
open new opportunities for the development of intelligent radar information processing methods. Promising
areas of further research include the development of adaptive neural network training methods in a
changing radio environment and the creation of hybrid systems combining traditional signal processing
methods with modern neural network algorithms