In the spirit of innovation, the development of an intelligent robot system incorporating the basic principles of Industry 4.0 was one of the objectives of this study. With this aim, an experimental application of an industrial robot unit in its own isolated environment was carried out using neural networks. In this paper, we describe one possible application of deep learning in an Industry 4.0 environment for robotic units. The image datasets required for learning were generated using data synthesis. There are significant benefits to the incorporation of this technology, as old machines can be smartened and made more efficient without additional costs. As an area of application, we present the preparation of a robot unit which at the time it was originally produced and commissioned was not capable of using machine learning technology for object-detection purposes. The results for different scenarios are presented and an overview of similar research topics on neural networks is provided. A method for synthetizing datasets of any size is described in detail. Specifically, the working domain of a given robot unit, a possible solution to compatibility issues and the learning of neural networks from 3D CAD models with rendered images will be discussed.

Industry 4.0 has become one of the most dominant research areas in industrial science today. Many industrial machinery units do not have modern standards that allow for the use of image analysis techniques in their commissioning. Intelligent material handling, sorting, and object recognition are not possible with the machinery we have. We therefore propose a novel deep learning approach for existing robotic devices that can be applied to future robots without modification. In the implementation, 3D CAD models of the PCB relay modules to be recognized are also designed for the implantation machine. Alternatively, we developed and manufactured parts for the assembly of aluminum profiles using FDM 3D printing technology, specifically for sorting purposes. We also apply deep learning algorithms based on the 3D CAD models to generate a dataset of objects for categorization using CGI rendering. We generate two datasets and apply image-to-image translation techniques to train deep learning algorithms. The synthesis achieved sufficient information content and quality in the synthesized images to train deep learning algorithms efficiently with them. As a result, we propose a dataset translation method that is suitable for situations in which regenerating the original dataset can be challenging. The results obtained are analyzed and evaluated for the dataset.

Electric vehicles are playing an important role in transport, aided by rapid advances in battery technology. The Faculty of Engineering at the University of Debrecen is also engaged research and development in the field of electric vehicles. To support the development of electric vehicle prototypes, a vehicle dynamics simulation program has been designed. The study presents the modeling and simulation of a permanent magnet synchronous motor (PMSM) in MATLAB/Simulink, which has been integrated into the existing vehicle dynamics simulation framework. The methods used to determine the motor characteristics required for the simulation are described in detail. In addition, the performance of the PMSM is compared with that of a brushless DC (BLDC) motor within the vehicle dynamics simulation program. The developed method allows the selection of the appropriate motor type for the given competition tasks.

Older multi-axis industrial robots used in industry do not have the computational power to use neural networks to perform complex manufacturing tasks. Therefore, a method combining 3D CAD modeling with data synthesis has been developed. The pix2pixHD approach allows the synthesis of photo-realistic images to generate datasets. The resulting datasets can be used to train deep learning-based detectors, which can later be applied to improve the object detection accuracy of older machine units. After evaluation of tests with YOLOv3 models, it is shown that only the datasets generated using the synthesized approach can detect an object with high detection accuracy. The generated datasets were also measured using the Complex wavelet structural similarity. The developed innovative method can be a practical and cost-effective strategy for smaller laboratories and small and medium enterprises, thus providing an opportunity to upgrade older machines.

Sciendo provides publishing services and solutions to academic and professional organizations and individual authors. We publish journals, books, conference proceedings and a variety of other publications.

This paper presents a novel synthetic learning-based approach for solving the component-to-slot assignment problem in robotics using a SCARA robot. The method uses a fully simulated environment that generates and annotates scenes based on rules and visual features. Within this environment, we train a permutation-invariant neural model to predict correct assignments between detected components and predefined target slots. Set Transformer-based encoders are combined with a self-attention MLP scoring head. Assignment prediction is optimized using an improved soft Hungarian loss function. To increase data realism and generalizability, we implement a synthetic dataset generation module on the NVIDIA Omniverse platform. This setup enables precise control over scene composition and component placement. The resulting model achieves high matching accuracy on complex layouts with variable numbers of components and demonstrates strong generalization across multiple configurations. Our results validate the feasibility of learning bijective mappings in simulated assembly scenarios, providing a foundation for scalable real-world robotic pick-and-place tasks. Tests were also conducted on actual robot units.