Automation 2025

1.
Towards an autonomous system: flexible modular production system enhanced with large language model agents
Michael Weyrich
Universität Stuttgart, Germany

2.
Biodegradable Flexible Sensors: Properties and Possibilities
Vytautas Bučinskas
Vilnius Gediminas Technical University, Lithuania

3.
Lie bracket – the key to nonholonomic motion planning
Ignacy Dulęba
Wrocław University of Technology

4.
Autonomous Calibration of Machine Vision Systems for Label Inspection Using Artwork Files as the Reference Source for ROI – 1st worldwide
¹Marcin Malesa, ¹Krzysztof Malowany, ²Piotr Rajkiewicz, ²Marta Grzegrzułka
¹KSM Vision, PhD, ²KSM Vision, Master of Science
m.malesa@ksmvision.com, k.malowany@ksmvision.com, p.rajkiewicz@ksmvision.com, m.grzegrzulka@ksmvision.com
Abstract: The aim of the automated machine vision system developed under the R&D project is to automate and streamline the calibration of quality inspection systems for the quality control of the packaging labels (at the 1st stage based on the examples of jar and pouches’ labels). In the project we describe the developed algorithms for automatic ROI selection based on artwork file label defect classification. Current solutions on the market require manual handling, especially in terms of calibration and parameterization for new formats and products. Traditional calibration methods require the use of actual reference photos taken at the start of production, which is time-consuming and can lead to errors due to differences in lighting conditions, product positioning or the quality of the photo itself. Human errors also occur during manual handling of this process. The solution developed under the R&D project uses an innovative approach – off-line calibration based on an artwork file, which eliminates the need to physically collect reference photos and allows the format preparation process to be carried out in an office environment, without interfering with the actual production line. As part of the R&D project, an innovative approach to autocalibration of the jar and pouch label inspection system has been developed and implemented, which significantly improves the processes involved in positioning Regions of Interest (ROIs). A key feature of the presented solution is the ability to calibrate based on an artwork file, eliminating the need for actual reference images. Using technologies such as LabelStudio and the conversion of PDF files into artwork formats, a high degree of accuracy and flexibility in the positioning of ROIs has been achieved. The solution developed allows real images to be effectively mapped to artwork, allowing the best images to be dynamically typed for analysis and solving problems related to differences in resolution and relationships between images. While label defect classifiers can work similarly for both products, the calibration is more complex in the case of jars. An additional step is required to accurately stitch images from cameras around the jar. The solution presents an innovative approach to calibrating a 360-degree product inspection system with labels, significantly streamlining and automating the processes involved in positioning Regions of Interest (ROIs) on jar labels. A key element of this solution is the ability to calibrate the layout from an artwork file, without the need to use actual reference images taken during production start-up. During the project, other potential approaches, such as calculating relationships between straightened images and the stitching technique, were also explored and ultimately discarded in favour of a more reliable and efficient solution based on offline calibration with matching of images to the artwork file. Technical insights of this innovation are to be shared with the participants of the ‘automation 2025’ Conference.

5.
Balancing Timeliness and Resource Limitations in Vehicle Routing Problems with Time Windows
Grzegorz Radzki, Grzegorz Bocewicz, Zbigniew Banaszak
Koszalin University of Technology
grzegorz.radzki@tu.koszalin.pl, Grzegorz.Bocewicz@tu.koszalin.pl, zbigniew.banaszak@tu.koszalin.pl
Abstract: This paper addresses the Vehicle Routing Problem with Time Windows and Lim-ited Resources (VRP-TR), which involves managing routes to connect customers for the delivery and pickup of service teams using different vehicles at the begin-ning and end of specified time windows. The model assumes that service teams operate in a multimodal mode, allowing them to be delivered by one mode of transport and picked up by another. The problem is constrained by the size of the vehicle fleet and the number of available service teams, which together limit the ability to process orders within the required time frames. Acknowledging scenar-ios where not all customers can be served on time, the study seeks to develop service plans that balance the timely provision of services to clients with the costs incurred by the service provider. The implementation of this problem within a de-clarative modeling framework enabled computational experiments, the results of which demonstrate its applicability to practical scenarios such as planning home healthcare visits or home appliance service appointments.

6.
Scalable PLC-based transport system control
Andrzej Wojtulewicz
Politechnika Warszawska, EITI, IAiIS
andrwoj@gmail.com
Abstract: The essence of this work is to present a universal PLC software concept and functional HMI visualisation for a modular transport line solution. The line consists of a series of independent modular roller conveyors and a turntable module combined into a coherent whole. The main part of the work was the development of a motion algorithm, but other functionalities of standard industrial facilities also had to be taken into account. The aim was also to develop a version of the program so that further transport modules could be added relatively easily. It was also important to have a flexible approach to the task in order to allow the proposed control to be applied in similar industrial systems.

7.
Assessing the effectiveness and efficiency of selected solution approaches for two-dimensional stock cutting problems (Part III): Hybrid Approach For Printed Circuit Boards
¹Stefan Bracke, ¹Leon Hoffe, ¹Kathrin Klamroth, ²Berna Ulutas
¹University of Wuppertal, ²Eskisehir Osmangazi University
bracke@uni-wuppertal.de, leon.hoffe@uni-wuppertal.de, klamroth@math.uni-wuppertal.de, bhaktan@ogu.edu.tr
Abstract: In industrial production, the cutting problem is prevalent across many technical fields. The two-dimensional stock cutting problem (2DCSP) aims to minimize the scrap produced when cutting flat items from larger flat shapes. In this paper, we consider the practically relevant special case that multiple copies of the same rectangular items have to be cut out of larger rectangular stocks. This situation is relevant in the production of printed circuit boards (PCB). As the third part of a comprehensive research about different solution approaches for 2DCSP, this paper presents an algorithm for a hybrid approach that combines rulebased and mixed integer linear programming methods. While generally applicable, the method is tailored to the specific situation considered in this paper and takes advantage of the problem structure. The results of the hybrid approach are discussed in terms of practical application and are compared to the rule-based approach.

8.
Implementation of automated measuring and robotic systems for quality control and economic challenges of small-scale manufacturing in Ethiopia
Wasihun Keno, Janusz Szpytko
AGH University of Krakow
wasihun@agh.edu.pl, szpytko@agh.edu.pl
Abstract: Most of the manufacturing industries across all over the globe are struggling to meet the increasing demands for precision, efficiency, and cost-effectiveness nowadays. However, small-scale manufacturing sectors in Ethiopia are facing significant economic challenges, such as low product quality, human labor shortages and rising production costs. To solve these challenges, automated measuring and robotic systems are a vital solution to improve product quality while also addressing the economic challenges and to make the products meet high quality standards by reducing material waste and increasing the productivity, which in turn makes to maintain competi-tiveness in the market. This paper examines the implementation of automat-ed measurement and robotic systems in small-scale manufacturing sectors in Ethiopia, emphasizing their role in improving quality control while address-ing economic obstacles that the country is facing today. The results show that automation can improve operational efficiency with reduced waste and in-creased product quality in local and international markets.

9.
A Review of Predictive Control Algorithms for Mobile Robots in advancing a trajectory tracking: Challenges, Opportunities, and Future Directions
¹Robel Hayelom Asgedom, ²Igor Korobiichuk
¹Independent Researcher, ²Łukasiewicz Research Network – Industrial Research Institute for Automation and Measurements PIAP,
robelasgedom629@gmail.com, igor.korobiichuk@piap.lukasiewicz.gov.pl
Abstract: Predictive control has become a critical methodology for enhancing the autonomy, adaptability, and precision of mobile robots, particularly in trajectory tracking applications. By anticipating future system behaviors, predictive control algorithms optimize performance while adhering to system constraints, enabling effective operation in dynamic and unpredictable environments. This paper provides a comprehensive review of state-of-the-art predictive control algorithms, including Model Predictive Control (MPC), Reinforcement Learning-based Predictive Control, and Adaptive Predictive Control, analyzing their advantages, limitations, and applicability to robotic trajectory tracking tasks. Key challenges, such as computational inefficiency, sensitivity to disturbances, and scalability to real-time applications, are identified as barriers to wider adoption. To address these is-sues, prospective research directions are proposed, focusing on integrating artificial intelligence and adaptive strategies to improve computational efficiency and robustness. Future contributions aim to develop hybrid algorithms and validate them experimentally across diverse robotic platforms and terrains, ensuring practical applicability. By advancing predictive control strategies, this research seeks to support the development of robust and adaptable trajectory tracking solutions for wheeled mobile robots, with implications for autonomous vehicles, industrial automation, and military application.

10.
Extension of the Floquet-Lyapunov transformation to fractional linear systems with periodic time-varying parameters and descriptor fractional systems reduced to standard ones
Tadeusz Kaczorek
Bialystok University of Technology
t.kaczorek@pb.edu.pl
Abstract: The fractional linear systems with periodic time varying parameters are considered. The Floquet-Lapunov theory is extended to the fractional linear systems with periodic parameters. Sufficient conditions are established under which the descriptor fractional linear systems can be reduced to standard fractional linear systems. A procedure is given for the reduction of the descriptor systems and illustrated by the simple numerical example.

11.
The Floquet-Lyapunov transformation of Roesser discrete-time linear model with periodic coefficients
Lukasz Sajewski, Tadeusz Kaczorek
Bialystok University of Technology
l.sajewski@pb.edu.pl, t.kaczorek@pb.edu.pl
Abstract: The Floquet-Lyapunov transformation is extended to Roesser discrete-time linear model with periodic coefficients. A procedure for computation of the transformation is proposed and illustrated by a numerical example. This approach is extended to the first and second Fornasini-Marchesini models.

12.
Numerical Analysis of the Discrete, Fractional Order PID Controller using FOBD and CFE Approximations
Krzysztof Oprzędkiewicz
AGH Unversity
kop@agh.edu.pl
Abstract: This paper presents the numerical analysis of the discrete, approximated Fractional Order PID Controller (FOPID). The fractional parts of the controller are approximated with the use of the most known methods: Fractional Order Backward Dierence (FOBD) and Continuous Fraction Expansion (CFE). CFE is simplier and faster than the FOBD method, but its accuracy is not always satisfying. For both approximations optimimum sample time was found by minimizing of the cost function Integral Absolute Error (IAE). Additionally, to optimize of CFE its parameter a was applied. Results of numerical tests show that the FOPID using FOBD is more accurate in the sense of IAE cost function for FOPI and FOPID controllers, but CFE is more accurate for FOPD controller. Next, the FOBD requires to use of smaller sample time to obtain of good accuracy than CFE. This allows to conclude that FOPD controller using CFE can be applied in time crictical applications at bounded platforms, for example in robotics or numerical control.

13.
Simulation of cascade control for Two Rotor Aerodynamical System using FOPID and PID controllers
Jakub Żegleń-Włodarczyk
AGH University of Krakow
zeglen@agh.edu.pl
Abstract: Two Rotor Aerodynamical System is a non-linear system in which cross-coupling occurs. In this specific case the individual rotors of the system affect both the system angles: azimuth and pitch. Therefore, both values must be used in each feedback loop. This makes it possible to use cascade control. It was decided to use FOPID as superior controllers and PID as inferior. For comparison purposes, an analogous control system with only PIDs was prepared. The coefficients were determined using the GWO algorithm. All simulations were performed in the Matlab / Simulink environment. In addition to comparing the values of the cost function, the execution time of individual controllers was also checked.

14.
Using Predictive Models to Improve the Accuracy of Dough Process Control
¹Nataliia Zaiets, ²Nataliia Lutska, ¹Lidiia Vlasenko
¹National University of Life and Environmental Sciences of Ukraine, ²National University of Food Technologies
z-n@nubip.edu.ua, lutskanm2017@gmail.com, vlasenko.lidia1@gmail.com
Abstract: Machine learning helps to build predictive models to identify possible devia-tions and is the basis for innovative solutions for adaptive and predictive process control. The paper examines examples of using various machine learning methods to optimize the dough preparation process, and assesses their impact on the quality of the final product and the efficiency of the en-tire production process. The sponge method of pre-dough preparation in-volves two-stage fermentation and is one of the traditional methods in the baking industry, which ensures a rich taste and high-quality texture of bread. The main variables of the dough preparation process have been expertly es-tablished, based on which a model for monitoring and optimizing the main variables of the dough preparation process, which are target ones, is formed. The dough preparation process includes determining the optimal values of the initial variables characterizing the final quality of the dough. To achieve target values, models are used that predict them based on the current values of the input variables, including control and disturbing variables. Regression tree models were developed to predict various parameters, which made it possible to accurately determine the relationship between input and output variables. The patternsearch method has shown high efficiency in problems with non-smooth functions, namely, visible phenomena character-istic of real technological processes. Having indicated the need to use local minimum values and exclude stability to achieve optimal indicators, the method has shown positive results at the stages of training and testing mod-els. As a result, optimization of the technological parameters of the dough preparation model for an additional direct search method has been achieved, and significant improvements in the accuracy of the automated process con-trol system have been achieved.

15.
Automation of classification and prediction in nephrological practice: a design for an expert system – preliminary research
¹Dawid Pawuś, ¹Szczepan Paszkiel, ²Tomasz Porażko
¹Opole University of Technology, Faculty of Electrical Engineering, Automatic Control and Informatics, ²Department of Internal Medicine and Nephrology, Institute of Medical Sciences, University of Opole
dawid.pawus@student.po.edu.pl, s.paszkiel@po.edu.pl, tomasz.porazko@usk.opole.pl
Abstract: This study explores the development of a modular expert system designed to automate key aspects of nephrological care, focusing on prediction, classification, and decision support. The research covers initial results from two implemented modules—predicting laboratory test results using machine learning and supporting risk classification in idiopathic membranous nephropathy. Machine learning models such as the adaptive input-output predictive model demonstrated high performance in predicting laboratory results with high accuracy. The risk classification system for idiopathic membranous nephropathy used rule-based expert algorithms and machine learning to improve clinical decision making, achieving decision accuracy superior to historical clinical decisions. These preliminary findings highlight the potential of AI to enhance diagnostic accuracy, streamline patient care, and improve outcomes in nephrology.

16.
STUDY OF ULTRASONIC DESTRUCTION OF MICROALGAE USING A CASCADE EMISSION PLANT
¹Vladyslav Shybetskyi, ²Igor Korobiichuk, ³Sergii Kostyk, ³Oleksii Kolisnichenko, ³Myroslava Kalinina
¹Łukasiewicz Research Network – Industrial Research Institute for Automation and Meas-urements PIAP, ²Warsaw University of Technology, ³National Technical University of Ukraine, Igor Sikorsky Kyiv Polytechnic Institute
vladyslav.shybetskyi@piap.lukasiewicz.gov.pl, igor.korobiichuk@piap.lukasiewicz.gov.pl, kostyk.serhii@lll.kpi.ua, Kaa2001@ukr.net, kalinina.kpi@gmail.com
Abstract: This study examines the ultrasonic disintegration of microalgae cells using a cascade emission plant, focusing on Chlorella cultures. A computer simula-tion was conducted to model pressure distribution, shear stresses, and cavita-tion effects within the device. The simulation results revealed maximum acoustic pressure of 3416 Pa and the highest surface deformation along the Y-axis at 2.863 × 10⁻⁹ m, confirming the validity of the device’s operational frequency and design. Experimental research investigated the effects of ultrasonic power, irradia-tion time, and suspension concentration on cell disintegration. The highest cell reduction (68.24%) was observed at a 1:1 concentration and maximum power after 20 minutes, with significant reductions occurring within the first 3-5 minutes. At a 1:3 concentration, cell reduction reached 52.17%, demon-strating a concentration-dependent ultrasound efficiency. Experiments with-out liquid circulation indicated intensified cavitation, achieving 63.27% re-duction at a 1:3 concentration. These findings highlight the effectiveness of ultrasonic technology in en-hancing microalgae disintegration for biofuel and lipid extraction, providing a foundation for optimizing ultrasonic systems in industrial applications.

17.
Modeling of ultrasonic emitter surface oscillations for ultrasound enhanced processes
¹Vladyslav Shybetskyi, ¹Igor Korobiichuk, ²Myroslava Kalinina, ²Daryna Khyzhna, ²Zlata Shopova
¹Łukasiewicz Research Network – Industrial Research Institute for Automation and Meas-urements PIAP, ²National Technical University of Ukraine, Igor Sikorsky Kyiv Polytechnic Institute
vladyslav.shybetskyi@piap.lukasiewicz.gov.pl, igor.korobiichuk@piap.lukasiewicz.gov.pl, kalinina.kpi@gmail.com, dariakhyzhna@gmail.com, zsshopova@gmail.com
Abstract: Ultrasonic systems are widely used for enhancing chemical, physical, and environmental processes, including particle coagulation, cavitation, and acoustophoresis. However, accurately modeling the motion of an emitter’s surface remains a challenge due to its complex interaction with surrounding media. This research addresses this issue by developing a method that con-siders directional displacements of all emitter surfaces in contact with the fluid across three axes. This study explores a novel approach for modeling ultrasonic emitter surface oscillations in computational fluid dynamics. The methodology involves determining natural frequencies and modal shapes us-ing Modal analysis, followed by simulating directional surface displacements in Harmonic Acoustic analysis. The approach improves the accuracy of pressure distribution and flow field predictions by incorporating realistic boundary conditions, such as sinusoidal displacement profiles and dynamic mesh approach. These findings are validated against experimental data, con-firming the model’s reliability for predicting fluid behavior under ultrasonic excitation.

18.
The sensor fault tolerance estimation scheme by reduction model for strongly-coupled systems
¹Viacheslav Ivashchuk, ²Igor Korobiichuk
¹National University of Life and Environmental Sciences of Ukraine, ²Łukasiewicz Research Network – Industrial Research Institute for Automation and Meas-urements PIAP
v.ivaschuk@nubip.edu.ua, igor.korobiichuk@pw.edu.pl
Abstract: In this paper, the algorithm of a fault-tolerant estimation that is capable of handling sensor full faults in addition to modelling uncertainties had been represented. The represented implicit approach allows estimation state for some smaller tolerance without the need for accurate information from the faulted sensor. In most cases, the only part of system is measured. In that way, the researching dominant functions of processes allows for deeper understanding of data provenance. Obviously, the observable nonlinear systems are not diffeomorphic to linear systems up to output actions and inputs of those systems may do not distinguish some number of different initial states. Based on some practical estimations for partly dependent variables, we provide empirical guarantees on the recovery rate of the sparse variables and the identification of the trusty example functions for the corresponding problem. So that makes demand to the complexity of the singular inputs analysis, although a complete theory which allows the design of an observer for these unobservable systems does not exist. Computational results are demonstrated on datasets generated by the technological object.

19.
Applicability of measuring the upper extremity’s range of motion with 3d-printed passive exoskeleton
¹Jan Oleksiuk, ¹Piotr Falkowski, ²Mateusz Zadrożniak, ²Julia Wilk
¹Lukasiewicz Research Network – Industrial Research Institute for Automation and Measurements PIAP, ²Warsaw University of Technology
jan.oleksiuk@piap.lukasiewicz.gov.pl, piotr.falkowski@piap.lukasiewicz.gov.pl, mateusz.zadrozniak@gmail.com, julia.wilk.kontakt@outlook.com
Abstract: Implementing exoskeleton robots for remote motor rehabili- tation offers a solution for the increasing demand for an ageing and grow- ing society, providing effective and consistent treatment service, reducing therapist workload, and potentially being more cost-efficient, especially up against pandemic reality. This study examined a validation process of the applicability of the upper extremity’s range of motion measurement with the custom-designed, 3D-printed, passive 5-DOF exoskeleton sup- plied with optical rotary encoders. Four healthy participants performed 2 exercises involving complex shoulder and elbow joint motion while wearing the exoskeleton. Data obtained from the exoskeleton was com- pared to data from a Motion Capture System (MSC) used in the previous study. Significant dissimilarities in mean values of achieved ranges, veloc- ities, and accelerations were observed for various joint angles; however, satisfactory results were obtained at plotted scopes levels. The study discussed possible reasons for differences in joint angle measurements between the exoskeleton and MSC methodologies. It highlighted obsta- cles related to the construction of the exoskeleton and differences in participant movement patterns.

20.
Adaptation of a large-scale, modular printer-milling machine for 3D printing from concrete
Wojciech Kiński, Przemysław Siwicki, Jacek Zieliński, Michał Smater
Łukasiewicz Research Network – Industrial Institute of Automation and Measurements PIAP
wojciech.kinski@piap.lukasiewicz.gov.pl, przemyslaw.siwicki@piap.lukasiewicz.gov.pl, jacek.zielinsk@piap.lukasiewicz.gov.pl, michal.smater@piap.lukasiewicz.gov.pl
Abstract: This article concerns the adaptation of a modular, large-scale printer-milling machine for printing from concrete, in order to be able to use it in automation in construction. The Łukasiewicz-PIAP research team, which includes the authors of this article, proposes the adaptation of the existing structure for 3D printing from concrete. The solution described in this article is the reconstruction of the executive tool and the appropriate software of the subassembly together with the selection of parameters of the manufacturing process. As a result of the work, an example of the use of the rebuilt structure was shown: printing full-size concrete stairs.

21.
Automated production line process validation using RFID technology
¹Arkadiusz Stęchły, ¹Bartosz Pawłowicz, ²Marek Ruman, ¹Bartosz Trybus, ¹Adrian Kosior
¹Politechnika Rzeszowska, Wydział Elektrotechniki i Informatyki, ul. Wincentego Pola 2, 35-021 Rzeszów, ²Elcont sp.z o.o. Trzebownisko 928c 36-001 Trzebownisko
a.stechly@prz.edu.pl, barpaw@prz.edu.pl, mruman@elcont.pl, btrybus@prz.edu.pl, adrian.kosior@kneiti.prz.edu.pl
Abstract: RFID technology allows for the automation of production processes, especially in quality control, resource management and production optimization roles. In this study, a laboratory model of a production line was created, using RFID tags to identify the accurate processing of a component at several processing stages and its final sorting. Test results confirm the effectiveness of the system in product classification and show the influence of orientation and distance between tags and readers on system performance.

22.
Effectiveness of object identification in ware-houses using RFID
¹Karol Siwiec, ¹Michał Ćmil, ¹Andrzej Paszkiewicz, ²Bartosz Pawłowicz, ³Łukasz Pachla
¹Zakład Systemów Złożonych, Politechnika Rzeszowska, ²Zakład Systemów Elektronicznych i Telekomunikacyjnych, Politechnika Rzeszowska, ³Wydział Elektrotechniki i Informatyki, Politechnika Rzeszowska
k.siwiec@prz.edu.pl, m.cmil@prz.edu.pl, andrzej.paszkiewicz@prz.edu.pl, barpaw@prz.edu.pl, 177135@stud.prz.edu.pl
Abstract: RFID technology has a crucial role to fulfill in the automation of logistics processes, providing fast and contactless identifi-cation of objects in distributed warehouse environments. The article examines the concept of research on the effectiveness of this technology, focusing on the reading speed and effi-ciency of RFID readers themselves under different operational conditions. Most attention was granted to the performance of RFID systems dependent on both device design and envi-ronmental factors such as humidity, tag obscuration and antenna placement. Preliminary measurements confirm that optimal RFID implementation can significantly increase the efficiency of inventory management and simultaneously minimize identification errors, making the technology indis-pensable in logistics and warehouse management.

23.
Automatic temperature control system in the iron based on RFID technology
Marcin Goral, Bartosz Pawłowicz, Paweł Wróbel, Nikita Bailiuk
Politechnika Rzeszowska, Wydział Elektrotechniki i Informatyki, ul. Wincentego Pola 2, 35-021 Rzeszów
m.goral@prz.edu.pl, barpaw@prz.edu.pl, 164130@o365.prz.edu.pl, n.bailiuk@prz.edu.pl
Abstract: The article showcases an innovative design for an RFID-integrated iron, which is part of the broader concept of the Internet of Textile Things (IoTT). The goal of the project was to develop a device capable of automatically reading information from RFID tags sewn into fabrics and dynamically adjusting the ironing temperature according to the type of fabric. The paper describes the design of the device, including the PCB design, the heating power control circuit and the solutions used to minimize electromagnetic interference.

24.
Implementation of IEC 61131-3 Functional Blocks Using PIO State Machines
¹Marcin Hubacz, ²Bartosz Trybus, ¹Bartosz Pawłowicz
¹Rzeszow University of Technology, ²Rzeszów University of Technology
m.hubacz@prz.edu.pl, btrybus@kia.prz.edu.pl, barpaw@prz.edu.pl
Abstract: The paper discusses the implementation of functional blocks compliant with the IEC 61131-3 standard using PIO (Programmable Input/Output) state machines available in systems such as the RP2040. A hybrid controller architecture is proposed, where the main control algorithm is executed in IEC 61131-3 languages via a virtual machine, but with support from native PIO blocks. Various implementations of RS, SR, and CTD blocks using PIO were analyzed, comparing their execution time with that of an implementation in the ST language. It was demonstrated that native PIO blocks are significantly faster than those executed by the virtual machine. Despite some limitations, such as the need to pre-program blocks in firmware or the limited number of instructions in PIO blocks, this solution offers significant benefits in applications where the execution time of operations is critical.

25.
Tabu Search Algorithm for Communication Time Optimization in the CPDev Engineering Environment
Andrzej Bożek, Dariusz Rzonca
Rzeszow University of Technology
abozek@kia.prz.edu.pl, drzonca@kia.prz.edu.pl
Abstract: Efficient communication is crucial for the correct operation of distributed control systems. This article is focused on the problem of reducing the communication cycle time during reading mixed register-binary data in the Modbus protocol from an industrial controller, especially implementing the CPDev engineering environnment. A tabu-search algorithm was designed and implemented for optimization. Extensive computational experiments were performed and their results were compared with the results of previous studies in which constraint programming was used for the same problem. Comparison of the results revealed that the new algorithm is much faster and therefore suitable for practical use in real-time Modbus communication planning for the CPDev environment.

26.
AN INTEGRATED ROBOTIC SYSTEM: EXPERIMENTAL VALIDATION OF FAULT DETECTION ROBOT FOR OVERHEAD CRANES
¹Arun Kumar Yadav, ²Janusz Szpytko
¹AGH University of Science & Technology, ²AGH University of Krakow
yadav@agh.edu.pl, szpytko@agh.edu.pl
Abstract: The purpose of this paper is to propose a new model of magnetic wheeled climbing robot for climbing various types of structures that include significant obstacles, besides inspection of industrial vessels made of various materials, including non-ferromagnetic material. The inspection consists of live video streaming of the inspection surface, sensing the surface and sub-surface cracks by using the GMR sensor array. The design mainly consists a flexible structure of the mobile robot with four specially designed magnetic wheels. An improved magnetic adhesion way by using the magnetic wheels was used to as the adhesion system for this climbing robot. This system generates the required adhesion force to support the robot on the climbed surfaces. Dynamic simulation and calculation of magnetic flux density, using ADAMS and FEM are performed and ensures the success of this idea. Experimental tests to check the capabilities of inspection robot, climbing different surfaces, such as smooth, rough, flat and cylindrical surfaces like the real vessel, are successfully carried out. In addition, the robot stops accurately on the climbed surface at any desired location for inspection purposes, and it overcomes significant obstacles up to 40 mm. This proposed climbing can be implanted for industries such as petrochemical, construction, bridges and many more where a regular inspection of the welds and the wall thickness is required. But here the primary motive task is the inspection of overhead cranes girders for which this robot is proposed.

27.
Industrial robot toolpath generation using Ramer-Douglas-Peucker method with arc recognition
¹Mantas Makulavicius, ²Kira Kiruta, ¹Ugnė Piliukaitytė, ¹Justas Makutėnas, ³Andrius Dzedzickis
¹Vilnius Gediminas Technical University, ²Klaipeda University, ³Vilnius Gediminas technical university
m.makulavicius@vilniustech.lt, kira.kiruta@ku.lt, ugne.piliukaityte@stud.vilniustech.lt, justas.makutenas@vilniustech.lt, andrius.dzedzickis@vilniustech.lt
Abstract: The efficiency of robot control methods and the accuracy of their move-ments are often hindered by the complexity of defining motion paths. Com-monly, these paths are represented by numerous linear segments, resulting in redundant motion commands, reduced smoothness, and velocity. To address these issues, this study investigates the application of mathematical algo-rithms, specifically the Ramer-Douglas-Peucker (RDP) method and higher-order polynomial interpolation, to optimize robot toolpaths. The RDP algo-rithm minimizes redundant points while retaining the trajectory’s overall shape, enhancing process speed and reducing energy costs. In addition, arc recognition is implemented to replace multiple linear operations with circular or arc-based motions, improving efficiency and reducing code complexity. The research employs RoboDK and Python API for offline programming, with experimental validation conducted on the Yaskawa Motoman HC10DTP robot. Results demonstrate that these methods can significantly reduce toolpath execution times – up to 3.7 times in certain cases – without significantly compromising trajectory accuracy. These approaches are appli-cable across various industries, including milling, painting, welding, and grinding, where smooth and precise motion is critical.

28.
Vulnerability detection in mobile robot odometry
Daniel Giełdowski, Wojciech Szynkiewicz
Warsaw University of Technology
daniel.gieldowski.dokt@pw.edu.pl, Wojciech.Szynkiewicz@pw.edu.pl
Abstract: Cyber-security is a critical issue for social and service robots. Due to their direct interaction with humans, it is essential to implement adequate security measures. This paper addresses the challenge of detecting vulnerabilities in odometry systems caused by cyber-attacks on a robot’s onboard sensors. Odometry is the primary method of relative localization used in autonomous robots. The proposed approach involves automatically detecting anomalous robot behaviour that may result from a cyber-attack. We employ an autoencoder trained on data collected during normal robot operation to identify such anomalies. Experimental results with a real robot demonstrate the effectiveness of the proposed method in detecting vulnerabilities in the robot’s odometry.

29.
DEFINITION AND ANALYSIS OF THE INDICATOR OF THE DEGREE OF RISK TO THE SAFETY OF AN AIRCRAFT FLIGHT ALONG THE OPTIMAL TRAJECTORY FOR AVOIDING MOVING OBSTACLES
Jerzy Graffstein
Łukasiewicz Institute of Aviation
jerzy.graffstein@ilot.edu.pl
Abstract: Ensuring safe avoidance of collisions with moving obstacles during flight requires carrying out several-stage activities. Their scope includes detecting obstacles, identifying the possibilities of potential collisions, calculating a safe flight trajectory, which is aimed at avoiding obstacles. It is important to perform an analysis of the calculated trajectory in terms of its safe implementation. The paper presents the method of performing the analysis of the safety of avoiding obstacles using an indicator. The method of formulating this indicator and its use at the stage of preparing and implementing the trajectory of avoiding obstacles is discussed. The calculations of the trajectory were carried out by solving the problem of particle swarm optimization (PSO). The safety of the example maneuvers obtained in this way was described by the calculated courses of the safety risk level indicator.

30.
Design of a Human-Assistance Robot System with Contextual Action Recognition
Amanuel Ergogo, Teresa Zielińska
Warsaw University of Technology
amanuel.ergogo@gmail.com, Teresa.Zielinska@pw.edu.pl
Abstract: This paper presents a conceptual design for a proactive human-assisting robot system capable of recognizing human activities and responding proactively. The system leverages contextual human activity recognition to interpret human actions across diverse contexts, while behavior trees are utilized to define dynamic and interpretable robot behaviors. We outline the system architecture, incorporating contextual human action recognition (HAR), behavior trees (BTs), and ROS, using the Spot robot platform as a representative example. We explain how HAR enables the robot to provide proactive assistance, discuss its limitations, and introduce methodologies for contextual HAR to address these limitations, thereby enhancing the robot’s decision-making in complex human activity scenarios.

31.
Machine Learning-Enhanced AR for Independent Lower Limb Rehabilitation
Andrzej Ziemnicki, Norbert Mostowski, Piotr Skrzypczynski
Poznan University of Technology
a-ziemnicki@wp.pl, norbert.mostowski@student.put.poznan.pl, piotr.skrzypczynski@put.poznan.pl
Abstract: This paper presents an innovative augmented reality system designed to support lower limb rehabilitation for neurological patients, such as stroke survivors. Using machine learning driven pose estimation algorithms and Unity-based software, the system enables patients to perform exercises autonomously without the need for external sensors or constant supervision. Relying solely on RGB camera data from AR glasses, the system offers interactive and engaging rehabilitation sessions that can be performed at home. Experimental results demonstrate feasibility of this cost-effective solution.

32.
A Reconfigurable Rehabilitation Exoskeleton for Exercising in Seated and Laying Position
Piotr Kołodziejski, Piotr Falkowski
Łukasiewicz Research Network – Industrial Research Institute for Automation and Measurements PIAP
piotr.kolodziejski@piap.lukasiewicz.gov.pl, piotr.falkowski@piap.lukasiewicz.gov.pl
Abstract: Most of the currently developed exoskeletons are design to use during singular activity or in just one out of the main positions for rehabilitation, which means that only part of the therapy can be supported by one device and the rest of it must either be a traditional one or involve other devices. Hence comes the need for a single robot to support multiple rehabilitation positions. The goal of this research was to create a frame that would allow for the exoskeleton to aid in both seated and laying positions during multiple stages of recovery and most exercises possible. Using an exoskeleton from polish research institute as an exemplary one and following antrophometrical criteria a series of designs were explored with significant focus on modularity and enabling multiple positions. All other tested approaches to the frame’s design were described in the article. FEM analysis of the final one was additionally conducted to ensure its durability.

33.
Analysis of anatomical compensations of activities of daily living for automatic exoskeleton-aided task-oriented physiotherapy
¹Piotr Falkowski, ¹Maciej Pikuliński, ²Tomasz Osiak, ²Kajetan Jeznach, ²Krzysztof Zawalski, ²Piotr Kołodziejski, ²Jan Oleksiuk, ²Andrzej Zakręcki, ³Daniel Śliż, ³Przemysław Kasiak, ⁴Natalia Osiak
¹ŁUKASIEWICZ Research Network – Industrial Research Institute of Automation and Measure-ments PIAP, ²ŁUKASIEWICZ Research Network – Industrial Research Institute of Automation and Measurements PIAP, ³Medical University of Warsaw, ⁴Z Miłości Do Kości
piotr.falkowski@piap.lukasiewicz.gov.pl, maciej.pikulinski.dokt@pw.edu.pl, tomasz.osiak@awf.edu.pl, kajetan.jeznach@piap.lukasiewicz.gov.pl, krzysztof.zawalski@piap.lukasiewicz.gov.pl, piotr.kolodziejski@piap.lukasiewicz.gov.pl, jan.oleksiuk@piap.lukasiewicz.gov.pl, zakrecki@agh.edu.pl, sliz.daniel@gmail.com, przemyslaw.kasiak1@gmail.com, kontakt@z-milosci-do-kosci.pl
Abstract: This study explores anatomical compensations in patients performing activities of daily living (ADLs) during exoskeleton-aided physiotherapy. Patients with neurological conditions often exhibit compensatory movements due to decreased muscular force and limited range of motion, potentially leading to incorrect motion patterns and uneven stress on the musculoskeletal system. By analysing kinematic data from 22 ADLs measured with inertial measurement units (IMUs), the research identifies typical compensatory behaviours assigned to modelled degrees of freedom (DOFs) of extremity. The results indicate that effective monitoring and correction of these compensations are essential for enhancing rehabilitation outcomes. The developed models will inform the SmartEx-Home exoskeleton’s control systems, enabling individualized motion support and promoting proper movement techniques for improved patient recovery and independence in daily activities.

34.
Use of artificial neural networks to detect minimal motion patterns to control assistive devices by patients with a high level of disability
¹Krzysztof Zawalski, ¹Piotr Kołodziejski, ²Piotr Falkowski
¹Industrial Research Institute for Automation and Measurements PIAP, ²ŁUKASIEWICZ Research Network – Industrial Research Institute of Automation and Measure-ments PIAP
krzysztof.zawalski@piap.lukasiewicz.gov.pl, piotr.kolodziejski@piap.lukasiewicz.gov.pl, piotr.falkowski@piap.lukasiewicz.gov.pl
Abstract: The increasing prevalence of individuals with limited mobility, particularly in ageing populations, necessitates the development of advanced assistive technologies. This study investigates the potential of using the measuring set based on Inertial Measurement Units (IMUs) combined with machine learning algorithms to detect minimal head movements for controlling assistive devices. A series of experiments were conducted using IMU sensors to capture and classify eight distinct head motions based on a registered dataset of approximately 11,000 samples. Various neural network architectures were evaluated, with the 5LSTM+2FC+ATT model achieving the highest accuracy of 85\% in motion classification. This research demonstrates that IMUs can effectively facilitate communication between patients with severe disabilities and assistive technologies, potentially enhancing their independence in daily activities. Future work will focus on refining sensor placement and improving model adaptability to individual user needs through personalized motion recognition.

35.
Vision-based automatic fruit counting with UAV
Hubert Szolc, Mateusz Wąsala, Remigiusz Mietła, Kacper Iwicki, Tomasz Kryjak
AGH University of Krakow
szolc@agh.edu.pl, wasala@agh.edu.pl, mietla@student.agh.edu.pl, kacperiwicki@student.agh.edu.pl, tomasz.kryjak@agh.edu.pl
Abstract: The use of unmanned aerial vehicles (UAVs) for smart agriculture is becoming increasingly popular. This is evidenced by recent scientific works, as well as the various competitions organised on this topic. Therefore, in this work we present a system for automatic fruit counting using UAVs. To detect them, our solution uses a vision algorithm that processes streams from an RGB camera and a depth sensor using classical image operations. Our system also allows the planning and execution of flight trajectories, taking into account the minimisation of flight time and distance covered. We tested the proposed solution in simulation and obtained an average score of 87.27/100 points from a total of 500 missions. We also submitted it to the UAV Competition organised as part of the ICUAS 2024 conference, where we achieved an average score of 84.83/100 points, placing 6th in a field of 23 teams and advancing to the finals.

36.
Mechatronic design and prototype of balancing vehicle
Filip Gomoracki, Robert Piotrowski
Gdańsk University of Technology
fgomoracki@gmail.com, robert.piotrowski@pg.edu.pl
Abstract: The aim of the article is to design and construct a mechanically and electrically self-balancing vehicle. The project was developed based on a series of technical assumptions and includes: selection of materials for the mechanical parts, design of custom components intended for production using 3D printing technology, preparing of a CAD model, selection of electronic components, design of the electronic system, and development of a remote control. Subsequently, a prototype of the vehicle was created, along with software enabling self-balancing and remote control via the controller.

37.
Characterization of Piezo-resistive Film Using Accurate Micro-Robotic System
Viktor Masalskyi, Jurate Jolanta Petroniene, Antanas Zinovicius, Marius Šumanas, Andrius Dzedzickis, Vytautas Bučinskas
Vilnius Gediminas Technical University, Vilnius LT-10223, Lithuania
viktor.masalskyi@vilniustech.lt, jurate-jolanta.petroniene@vilniustech.lt, antanas.zinovicius@vilniustech.lt, marius.sumanas@vilniustech.lt, andrius.dzedzickis@vilniustech.lt, vytautas.bucinskas@vilniustech.lt
Abstract: Recent growth in the widespread implementation of piezoresistive materials and their development raises a need for research on their characteristics, especially in the case of biopolymers. This research focuses on the defining of electrical characteristics of k-carrageenan-based piezoresistive film intended to be used as a transducer in force and pressure sensors. This film was prepared using the natural polysaccharide derived from Baltic seaweed Furcellaria lumbricalis; therefore, it is cheap and biodegradable. The main issue is material inhomogeneity due to natural reasons, surface microstructure, and various additives such as Fe2O3, which behave as a semiconductor. Characterization of the force-sensitive film was performed by using a high-accuracy micropositioning system combined with a resistance measurement device. The obtained results show the distribution of material resistance in the form of a map and support the hypothesis that real material surface area strongly differs from the geometric one, the film surface is not uniform, and has microcapillaries, therefore accurate definition of film surface charge transfer mechanism and its efficiency require experimental evaluation before implementing such film in practical measurement application.

38.
Seamless Indoor/Outdoor Transitioning in UWB/GNSS Positioning for Unmanned Ground Vehicles
¹Konrad Jurczyński-Chu, ²Arkadiusz Perski
¹Warsaw University of Technology, ²Łukasiewicz Research Network – Industrial Research Institute for Automation and Measurements PIAP
chukonrad@gmail.com, arkadiusz.perski@piap.lukasiewicz.gov.pl
Abstract: UWB (Ultra-Wide Band)- and GNSS (Global Navigation Satellite System)-based real-time positioning methods have been the subject of extensive research and perform well in their respective optimal environments. However, their optimal environments are largely mutually exclusive in practical applications; a cost-effective ubiquitous indoor/outdoor positioning system must utilize both systems to provide reliable positioning information in all environments. At the same time, there is no need to consider GNSS information in GNSS-denied areas (indoors) with UWB coverage. Therefore, a switch between the systems as the definitive positioning information source is needed. This paper outlines some approaches to transitioning between UWB and GNSS positioning of an unmanned ground vehicle (UGV).

39.
Extended Reality: Evaluating 3D scanning accuracy of Meta Quest 3
Oleksii Shatokhin, Vytautas Bučinskas, Sigitas Petkevičius, Nikolaj Šešok, Alfredas Kerulis, Antanas Zinovičius
Vilniaus Gedimino technikos universitetas
oleksii.shatokhin@vilniustech.lt, vytautas.bucinskas@vilniustech.lt, sigitas.petkevicius@vilniustech.lt, nikolaj.sesok@vilniustech.lt, alfredas.kerulis@vilniustech.lt, antanas.zinovicius@vilniustech.lt
Abstract: Rapid technological advances have made many high-tech devices available to the average user. Including Extended Reality headsets available for purchase for home use. Thanks to this, augmented reality technologies are finding their application in various fields from entertainment to professional use. XR, X Reality or Extended Reality is a general term meaning the technology of visualization and interaction with computer-generated reality. But these systems allow not only to visualize graphics, they also allow scanning a room and creating a 3D model of it, to prevent the player from colliding with walls and other obstacles. But the 3D scanning technology in these glasses can be applied in other, more professional fields, such as setting up and paving the way for mobile robotics, production planning and interior design sketches. This article describes the measurement of the accuracy of room scanning using the Meta Quest 3 mixed reality headset and describes possible improvements and applications.

40.
Algorithms for calculating navigation data of rotating missiles
Dawid Florczak, Szymon Ciszewski, Robert Głębocki
Faculty of Power and Aeronautical Engineering
01105138@pw.edu.pl, szymon.ciszewski.stud@pw.edu.pl, robert.glebocki@pw.edu.pl
Abstract: The key difficulty in navigating rotating missiles using inertial navigation systems (INS) equipped with MEMS sensors lies in the accurate measurement of their rotational velocity. Errors in this measurement are the primary contributors to inaccuracies in inertial navigation, posing a major challenge to improving and controlling traditional designs of unguided rotating objects. To overcome this issue, a novel system has been developed to aid the navigation of fast-rotating missiles. This study focuses on the development of a series of algorithms that calculate navigation data for rotating missiles by analyzing changes in the intensity of selected electromagnetic wave spectra. The system operates by employing a measurement setup that utilizes various sensors to record different types of electromagnetic radiation (visible light, infrared, and ultraviolet light). Navigation data estimation in the rotational channel will be performed using algorithms that calculate the time intervals between the recorded maximum and minimum values of the aforementioned radiation types. Ultimately, the goal is to develop algorithms capable of calculating navigation data based on measurements recorded under different atmospheric conditions, such as fog, heavy cloud cover, strong sunlight, and varying lighting conditions between day and night. It is anticipated that the most optimal algorithm for specific system conditions will be selected based on illumination and time-of-day factors.

41.
Dual-source bridges for measurement of two related quantities
Zygmunt Lech Warsza
Polish Merological Society
zlw1936@gmail.com
Abstract: The paper discusses basic relationships, an example of application and an outline of the state of research on the properties of unconventional measuring circuits with two current sources (2J) or two voltage sources (2E). Sources 2J are connected in parallel, 2E sources -in series to opposite branches of a four-impedance loop with two outputs from its diagonals. Formulas for output voltages and conditions of balance of both outputs of double-source direct current (DC) bridges are given. For double-current bridges, these are the equality of the products of resistances of the arms adjacent to the terminals of each output, and for double-voltage bridges – equality of the sums of arm resistances. These bridges, when both outputs are balanced, allow for the accurate measurement of resistance (conductance) values in their arms. The output signals depend on linear combinations of the increments of arm resistances from their values in equilibrium states. As an example of application, the simultaneous measurement of strain and temperature changes with one differential strain gauge is considered. The state of theoretical work and experimental research results of prototype designs of dual-current DC bridges and measuring devices and laboratory stands with their application are also shortly characterized. The state and directions of theoretical studies and experimental verification structures of measuring systems two-voltage DC bridges are briefly outlined. Two-source bridges of alternating current AC and of other waveforms have a huge variety of possible solutions for measuring two or more parameters. A bibliography of publications on dual-source bridges is also included.

42.
Statistical Reliability of Assessment Technological Process Compliance with Standards
¹Yevhen Volodarskyi, ²Oleh Kozyr, ³Zygmunt Warsza
¹National Technical University of Ukraine “KPI”, ²Igor Sikorsky Kyiv Polytechnic Institute KPI, ³Polish Metrological Socciety PTM
vet-1@ukr.net, olegkozyr86@gmail.com, zlw1936@gmail.com
Abstract: The article discusses the influence of probability distributions of characteristic parameters of a technological process and their measurement errors on the assessment of the compliance of this process with standards. Quantitative control cards are used for such an assessment of production processes. For this control, a priori probability of incorrect decisions is also estimated. In order to improve the statistical reliability of the process assessment, it is proposed to use a posteriori probabilities as well. As an example, the effectiveness of the assessment of the compliance of a technological process with standards is discussed when the probability distributions of its parameter values and their measurement errors are uniform.

43.
STATISTICAL ASSESMENT OF MIXED-REALITY SIMULATOR SERVER PERFORMANCE
¹Piotr Golański, ²Michał Golański
¹Instytut Techniczny Wojsk Lotniczych, ²Wojskowa Akademia Techniczna
piotr.golanski@itwl.pl, michal.golanski@itwl.pl
Abstract: Despite the continuous increase in the computing power of modern computers, their capabilities are still insufficient in many areas. An example is the simulation of physical processes taking place in all types of training systems based on Virtual or Mixed Reality technology. This is due to the growing requirements placed on these systems to achieve greater accuracy in reproducing reality. The solution to the above problem is the use of distributed simulation, which  by utilizing cloud solutions  allows for remote access to the computing power of specialized servers, and thus also provides the option of conducting a remote training. In such case, ensuring effective communication between the simulator’s client stations and the server in real time is of utmost importance. This article presents a proposal for a server-client communication protocol, for use in the cloud version of the Mi-17 helicopter, Mixed-Reality simulator. A method for evaluating performance of this protocol was proposed, along with a statistical description of the obtained results.

44.
A study of neural network efficiency in a binary classifi cation problem
¹Paweł Dymora, ²Mirosław Mazurek, ¹Przemysław Elgass
¹Faculty of Electrical and Computer Engineering, Rzeszów University of Technology, ²Politechnika Rzeszowska
pawel.dymora@prz.edu.pl, mirekmaz@prz.edu.pl, 169535@stud.prz.edu.pl
Abstract: This paper, explores the principles and applications of artificial intelligence, with a focus on CUDA technology and the PyTorch framework. The study provides an in-depth analysis of the neural network performance. Key topics include deep neural networks, activation functions, fundamental aspects of building artificial intelligence models, and practical applications of AI across various domains. Furthermore, the paper briefly compares alternatives to CUDA technology and answers the question of why CUDA would be a better option to use at the moment with PyTorch framework for machine learning purposes. By examining the interplay between hardware and software in modern AI workflows, the paper sheds light on the technological advancements driving innovation in the neural network training process. Additionally, the paper examines the growing significance of artificial intelligence in today’s world.

45.
Comparison of neural network training time using CPU and GPU with NVIDIA CUDA technology
¹Paweł Dymora, ²Mirosław Mazurek, ¹Przemysław Elgass
¹Faculty of Electrical and Computer Engineering, Rzeszów University of Technology, ²Politechnika Rzeszowska
pawel.dymora@prz.edu.pl, mirekmaz@prz.edu.pl, 169535@stud.prz.edu.pl
Abstract: The study investigates the impact of computational hardware on the training time of neural networks, comparing GPU and CPU performance in a binary classification problem. Performed experiments utilizing CUDA technology for GPU acceleration. The hardware specifications are thoroughly examined, covering local GPU and CPU configurations and cloud-based solutions, including Google Colab and Microsoft Azure. Experimental results compare the training times across local and cloud hardware. The findings demonstrate the significant advantages of GPUs in accelerating training, particularly for complex models and large datasets.

46.
Comparison of cardiovascular diseases’ classification models based on ECG signals using consumer-grade GPU
¹Kajetan Jeznach, ²Krzysztof Hryniów
¹Łukasiewicz Research Network – Industrial Research Institute for Automation and Measurements PIAP, ²Warsaw University of Technology
kajetan.jeznach@piap.lukasiewicz.gov.pl, krzysztof.hryniow@pw.edu.pl
Abstract: Deep neural networks (DNNs) are becoming a handy tool in the healthcare field. Research work in recent years has led to the creation of solutions that can effectively support the work of medical staff. This paper presents a comparison of deep learning architectures used for the classification of cardiovascular diseases. Models were trained and tested on the PTB-XL dataset, a large publicly available electrocardiography dataset containing electrocardiography (ECG) signals with disease labels assigned by qualified cardiologists. Some architectures achieved promising accuracy despite the poor computational capabilities of the consumer-grade GPU used for training DNNs. This study compares architectures, including LSTM, CNN, and GRU layers, inspired by the ones available in scientific publications. However, all of them were modified to be trained on a local machine, which has more computational limits than the ones used in professional AI field. The task to perform is multiclass & multi-label classification of five classes occurence (four cardiovascular diseases such as myocardial infarction, ST/T change, conduction disturbance, hypertrophy and one class for outcome of a healthy patient).