Theses (Department of Electrical Engineering and Mechatronics)
Állandó link (URI) ehhez a gyűjteményhez
Theses collection of the Department of Electrical Engineering and Mechatronics. The collection was started in 2023.
At the University of Debrecen, in accordance with the 2022 amendment to the 2011 Higher Education Act, student theses are only accessible from devices connected to the University's Eduroam WiFi network or from a university IP address.
“The thesis or diploma work of a student who has successfully passed the final examination shall be stored in full in the academic system of the higher education institution, and a record shall be maintained thereof. The stored theses and diploma works – with the exception of parts classified as confidential in accordance with the relevant legislation – must be made accessible and searchable without restriction through the academic system.” Further info on the National Higher Education Act in Hungarian: Felsőokt. tv. (új) - 2011. évi CCIV. törvény a nemzeti felsőoktatásról - Hatályos Jogszabályok Gyűjteménye.
Böngészés
legfrissebb feltöltések
Megjelenítve 1 - 20 (Összesen 134)
Tétel Korlátozottan hozzáférhető SIMULATION AND OPTIMIZATION OF V2G ENERGY EXCHANGE IN AN ENERGY COMMUNITYMansour, Abdelrahman Khaled Salah Abdelkader; Menyhárt, József; DE--Műszaki KarThis thesis shows how to improve energy exchange between electric vehicles and the grid in a small community using AI-based control. The community is made up of 50 EVs, a 150 kW PV system and real 2019 German load and solar data, simulated over a 24-hour period with 15-minute resolution. The thesis compares the most common optimization methods, NSGA-II, multi-objective PSO and a Deep Q, to demonstrate their ability to lower energy cost, reduce peak demand and limit battery degradation. A physics-based battery degradation model is included to ensure that the results reflect realistic battery behaviour. It was clearly shown in the thesis that the NSGA-II, among all other algorithms, gave the best and strongest overall performance, reaching to a 30% energy cost reduction while showing the best Pareto front hypervolume. The work shows that V2G becomes profitable when the time-of-use price spread is about 0.28 $/kWh or higher. Real OPSD profiles produced more variability but also more economic potential compared to synthetic ones.Tétel Korlátozottan hozzáférhető deep learning supported obstacle avoidance for universal ur10 collaborative robotic armambindi, benard; Korsoveczki, Gyula; DE--Műszaki KarThis Thesis presents a software framework that utilizes copper foil electrodes and an FDC2114 LC-resonant capacitance-to-digital converter to mathematically model the detection of conductive objects within a 30 cm radius of the cobot. The Soft Actor-Critic (SAC) DRL algorithm was trained through a four-stage curriculum within the CoppeliaSim simulation software, using a 61-dimensional state space that includes different measurements of the cobot’s sensors and its joint angles. A Control Barrier Function (CBF) prevents the cobot from entering a state that does not satisfy the safety constraint of remaining at a minimum distance to the detected obstacles, and a dual-loop Lagrangian method ensures the cobot reaches its goal while simultaneously satisfying the CBF constraint.Tétel Korlátozottan hozzáférhető Machine Learning Based Surface Hardnes and Vibration Strength EstimationYilmaz, Batuhan; Szilágyi, Péter; DE--Műszaki KarUsing machine learning a vibration based system was produced that can infer different surfaces from one another. The work includes supervised learning with captured samples of an accelerometer. Using testing samples never used in training, system was tested to identify 10 different surface types. Main concept includes fast, non-destructive and blackbox style architecture to perform detection. All relevant software, libraries, hardware and developed technique included.Tétel Korlátozottan hozzáférhető Obstacle Avoidance Using Hybrid Reinforcement Learning in Dynamic EnvironmentsPham, Tuan Nam; Almusawi, Husam Abdulkareem; DE--Műszaki KarThis thesis presents a hybrid reinforcement learning approach for obstacle avoidance with the Festo Robotino 4 omnidirectional mobile robot in dynamic environments. The proposed method combines a classical Pure Pursuit controller with a Soft Actor-Critic (SAC) agent that uses a custom 1D CNN (LidarCNN1D) to extract spatial features from stacked LiDAR observations. The system is implemented in ROS 2 Humble and Gazebo, and trained with a three-stage curriculum that scales from a 10×10 m room with static obstacles up to a 6×6 m room with six dynamic pedestrians. Four methods — Hybrid SAC, Optimized SAC, Pure SAC, and Pure PPO — are compared to isolate the contribution of each architectural choice. The Hybrid SAC achieves the best results, reaching a 99% success rate in the hardest training stage and 84% in the most difficult unseen evaluation environment, while maintaining the lowest collision rate (16%). The findings show that CNN-based feature extraction, frame stacking and Pure Pursuit guidance combine to deliver robust real-time navigation suitable for indoor service robots.Tétel Korlátozottan hozzáférhető Comparative Study of State-Space Control, Model Predictive Control and Sliding Mode Control for a Quarter-Car Active Suspension SystemMeng, Xianhao; Korsoveczki, Gyula; DE--Műszaki KarThis thesis compares three control methods for a quarter-vehicle active suspension system: state-space control, model predictive control, and sliding mode control. Dynamic models of the system are established using the Newton-Euler method, key-graph method, and Lagrange method, and the corresponding state-space equations are derived. The controllability and observability of the system are analyzed, including special cases such as chassis spring failure and tire stiffness failure. Then, three controllers are designed under the same physical parameters and tested under step and impulsive road disturbances. The study also considers variations in vehicle load to evaluate the robustness of the controllers. The performance of various methods is compared by rise time, overshoot, settling time, and steady-state error.Tétel Korlátozottan hozzáférhető Evaluation of Robot Suspension Systems, in a Human-Robot Interaction Point of ViewSamadli, Hasan; Mikuska, Róbert; DE--Műszaki KarI developed an active suspension system for mobile robots to enable them to communicate intent through fluid, biological body language, bridging the gap between functional robotics and social interaction. By modeling the robot using a quarter-car state-space approach and implementing it in ROS 2, I successfully created a system that mimics natural movement. I overcame technical challenges like kinematic ghosting by integrating IMU sensor data, which allowed me to optimize the damping ratio and achieve a critically damped, smooth response. To test this, I evaluated the system's stability under unpredictable conditions using a dynamic human tracking script in a Gazebo simulation. Ultimately, my research demonstrates that active suspension serves as a powerful non-verbal communication framework, providing robots with a natural presence that can significantly enhance user trust.Tétel Korlátozottan hozzáférhető Pruefcubing of the Future: Electrification of the CubeBen Youssef, Melek; Kis, Károly Árpád; DE--Műszaki KarElectrification of the BMW Group’s Pruefcubing process helped to overcome the limitation of using non-functional “dummy” control units for components such as the electrically extendable door handles that were included on several models of the Neue Klasse line. Through evaluating the feasibility of three different methods of integrating electrical components into the mock-up vehicles, the BMW Group determined that the Direct Node Actuation method was the most economical and autonomous means of creating the “Functional Islands” required within these static mock-ups. The implementation of an STM32F446RE microcontroller and an MCP2004A transceiver into the Pruefcubing mock-up vehicles permitted those static models to emulate the master control unit of the vehicle. The implementation of these components allowed the mock-up vehicles to operate their door handles in accordance with the specifications of the LIN 2.2 protocol, earning the implementation an “Outstanding” rating from the company. Overall, this implementation permitted for electrical and electronic (E/E) defects to be resolved three years prior to when the affected vehicles would go into production, reducing the costs that would otherwise be associated with correcting those defects during production.Tétel Korlátozottan hozzáférhető Design and kinematic analysis of a 4DoF delta-parallel mechanism using Onshape and MATLAB environmentBinth Ali Nasheed, Yooha; Korsoveczki, Gyula; DE--Műszaki KarThis thesis presents the design and Kinematic Analysis of a 4 DoF delta-parallel mechanism using Onshape and MATLAB environment. The purpose of the project was to design a linearly actuated 4 degrees of freedom delta parallel robot and to derive, implement and verify the kinematics of the robot. The intended purpose of the proposed 4 DoF delta robot is Pharmaceutical and laboratory applications. Including tasks such as working in controlled or hazardous environments with tasks such as pick and place, filling, capping and laboratory instrumentation where precision and accuracy are required. The thesis topic addresses the growing industrial demand for parallel robotic systems by providing a fully verified kinematic model and verification and visualization of the results. To create the robot first, the scope of application needed to be determined in order to finalize the size and dimensions of the robot. After this was realized the 2D Skelton was made and in accordance with the 2D skeleton diagram the 3D design was done on Onshape. After this the geometrical relationships between the parameters of the robot were realized, the Grübler–Kutzbach mobility formula was applied to confirm that the mechanism possesses exactly four degrees of freedom. To calculate the inverse and forward kinematics, mathematical calculations were derived and these calculations were transferred to MATLAB for computational verification along with visualization. A digital twin dashboard was developed integrating the verified solvers with a live three-dimensional skeleton rendering to provide interactive kinematic verification and visual documentation of the robot.Tétel Korlátozottan hozzáférhető Pruefcubing of the Future: Digital Twin Development for Automotive Measurement Integration in Pre-Series ProductionSami Rezk Abdelwahab Eldessouki, Ali; Huu, Tuan Nguyen; Károly Árpád, Kis; DE--Műszaki KarThis thesis details the development of a bidirectional Cyber-Physical Bridge for the BMW Group Center Measurement Room (CMR), integrating real-time physical data into a high-fidelity Unreal Engine 5 environment. By utilizing a Flask-based REST API and JSON protocols, the system transforms a static 3D model into a dynamic Digital Twin (DT) that reflects "as-manufactured" realities with sub-millimeter precision. A core technical contribution is the implementation of Atomic Write Integrity, which secures sensitive master records during data exchange and fosters user trust among engineering teams. Performance benchmarking confirms the system is industrially viable, achieving a stable 34.38 FPS and low-latency API response times of 8.7 ms on standard hardware. The framework addresses critical research gaps regarding the synchronization of heterogeneous measurement sources and the dependency on expensive physical prototypes. Furthermore, the virtual environment enables "fly-through" inspections of internal structural alignments that are physically inaccessible in the real world. Ultimately, this research provides a scalable foundation for the "Speed Cubing" initiative, aiming to accelerate automotive production cycles through enhanced virtual validation and human-system collaboration. This innovative approach demonstrates how Digital Twin technology can effectively replace resource-intensive physical procedures while maintaining the high accuracy required for automotive pre-series production.Tétel Korlátozottan hozzáférhető Developing a water leak sensor cable for smart home systemsShehata, Hussein; Szilágyi, Péter; DE--Műszaki KarThis summary progress report presents the development of a water leakage detection system for smart homes by using Raspberry Pi 5 and a homemade resistive sensing wire. The project’s main goal is to detect water leaks from pipes based on the change in resistance along the cable, so in the system I used an ADS1115 analog to digital converter to measure the voltage across the resistive sensing cable which changes when the leakages of the pipes occur and by using python in raspberry pi This project’s prototype is made for educational use within a mechatronics engineering context to demonstrate sensor interfacing, signal processing, and embedded system integration.Tétel Korlátozottan hozzáférhető Controller tuning of an industrial servo drive under variable conditionsBhari, Mohamed Amine; Mikuska, Robert; DE--Műszaki KarThis thesis presents the controller tuning and experimental evaluation of an industrial servo drive system under variable load conditions. The work is based on a Beckhoff AX5000 servo drive, an AM8022 servo motor, EtherCAT communication, and the TwinCAT 3 environment. The study investigates how tuning parameters such as proportional gain, integral time constant, and position gain affect velocity response, position tracking, stability, overshoot, oscillation, and tracking error. Experimental data were recorded with TwinCAT Scope View and evaluated using graphical comparison, RMS error, and peak-to-peak error. The results show that improper tuning causes vibration, oscillation, and larger tracking errors, while improved tuning leads to smoother motion and better accuracy. Overall, the thesis highlights the importance of systematic controller tuning and experimental validation in industrial motion control applications.Tétel Korlátozottan hozzáférhető Dynamic Simulation Robotic Arm Using MATLAB SimscapeJawad, Ali Nahyan; Nusser, Dávid Péter; DE--Műszaki KarThis thesis presents the dynamic simulation and PID-based control of a robotic arm using MATLAB Simulink and Simscape Multibody. The robotic arm is based on the ARX platform and is modelled as a four-active-joint serial manipulator, with the fifth joint treated as fixed. A SolidWorks CAD assembly was converted into URDF and STL files and imported into Simscape Multibody to create a virtual dynamic model. Denavit-Hartenberg parameters, forward kinematics, inverse kinematics, and trajectory planning were used to generate and verify the end-effector motion. The simulation evaluates joint torque, joint velocity, joint position, motor voltage, current, power, TCP position, and simplified TCP orientation through wrist compensation. The results show that the model can reproduce the planned robotic arm motion and provide useful information about actuator loading and controller performance within the assumptions of the simulation.Tétel Korlátozottan hozzáférhető SINGULARITY ANALYSIS OF THE 6 DOF FANUC INDUSTRIAL ROBOTIC ARMAli, Ola Abuelhussein Khamis; Gyula, Korsoveczki; DE--Műszaki KarThe thesis presents an analysis of singularity, decoupling techniques, and kinematics modeling for the FANUC LR Mate 200iD industrial robot. This articulated 6-DOF manipulator such as FANUC LR Mate 200iD is utilized in applications such as Machine tending, Inspection & Quality Control and assembling. However, it is inherently prone to singularity due to articulated structure and mathematics used to control. This leads to deteriorating the control and motion instability due to loss of rank in Jacobian matrix. Detecting these kinematic obstacles is essential for achieving operational stability, safety and efficiency. The research conducted is supported by visualization and simulation in CoppeliaSim using the Lua programming language. The FANUC LR Mate requires robust control strategies, including both forward kinematics (FK) and inverse kinematics (IK). The paper exhibits the role of Denavit-Hartenberg (DH) parameters and the decoupling method, which separates the position and orientation problems in inverse kinematics (IK) computation to analyze the complex arm .Furthermore, the study details the use of Jacobian matrix for validating the model, as well as examining and analyzing singularities that affect motion predictability and constraints. Moreover, the study delves into simulation using CoppeliaSim with Lua script integration to validate the Matlab’s kinematic model in the virtual world before real- world implementation. These simulations provide an effective means to visualize the robot’s workspace, trajectories and singular positions (such as elbow and arm singularities). The study managed to demonstrate the accuracy of singularity detection using condition number and Jacobian determinant in MATLAB, validating decoupling thorough Coppeliasim. Conclusively, this research demonstrates that the decoupling method is favored and direct technique for solving real-world control systems. It prevents errors and ambiguity, thereby helping us in understanding and monitoring of singularities in industrial systems for future incorporation with path optimization, vision control systems, and ROS.Tétel Embargó alatt Design and Kinematic Analysis of a Hexapod Stewart PlatformVelásquez Díaz, Claudia Valeria; Korsoveczki, Gyula; DE--Műszaki KarThis thesis develops an educational digital twin system for the Stewart Platform, a six-degree-of-freedom parallel manipulator that is rarely taught due to its complexity and the high cost of physical hardware. The system enables real-time online simulation of an educational-scale hexapod capable of reproducing motion profiles from aerospace and racing applications, allowing students to explore high-performance mechanisms without requiring access to advanced laboratories. The work guides learners through a complete development path (modeling, assembling, and simulating the platform from scratch), which strengthens understanding of kinematics, spatial reasoning, and robotic integration. By using accessible online environments, the thesis demonstrates that mechanics, electronics, and software function as a unified system in mechatronic design. The thesis promotes an exploration-based learning approach aimed at developing confident multidisciplinary engineers who understand complex robotic systems through hands-on experience and deep conceptual engagement.Tétel Embargó alatt mobile robot navigation algorithm with reinforcement learning method enhanced.Gmar, Ghofrane; Taleb abdullah abdo, Mayar; DE--Műszaki KarThis thesis studies the challenges of mobile navigation in a dynamic environment. In this thesi,s different categories of algorithms (graph-based, sample-based,...) are tested in static and dynamic environments to study their functionality, and then finally, I integrate reinforcement learning and potential field methods for mobile navigation enhancement. The final navigation approach is a hybrid approach that consists of a global planner( Hybrid A*and D*lite), a local planner( Q-learning and potential field method). The global planner provides the robot with a pre-defined path based on the given map and the local planner helps the robot to react in real-time to unexpected changes in the environment. The hybrid approach is a more stable and adaptable method, although it should be refined to ensure stability and consistency in performance.Tétel Korlátozottan hozzáférhető Optimization of Path Planning Algorithms for Autonomous Navigation in Simulated and Real EnvironmentsHakiri, Mohamed Amine; Husam , A. Neamah; DE--Műszaki KarThis works presents an in-depth performance comparison of path planning algorithms for autonomous mobile robot navigation under simulated and real environment. Five different navigation strategies were systematically contrasted: four planner combinations of A * and RRT* with DWA, TEB and an end-to-end (E2E) policy trained from scratch based on the Proximal Policy Optimization algorithm. The evaluation was performed on a home-made robot with differential drive kinematics, an RP LIDAR A1M8 laser scanner, an Arduino based module, and a Raspberry Pi 4 in indoor warehouse like scenarios. The experimental procedure consisted of several test situations using three different target locations in static and dynamic obstacle environments. Each configuration was executed three times for robustness, which is crucial in the context of probability-based RRT. Performance was assessed using multiple metrics including the number of calls to global planner representing replanning rates, time to reach goal, and for RRT – failed attempts at path generation. Experiments were done in simulation with Gazebo and the results was verified on the real robot. Simulation results indicated that A*-based combinations counted less replanning calls in comparison with RRT-based, and A+TEB, realized the best calculation time when operating in static environments. In dynamic obstacle environment, local planner decision played a more important role than the global planner selection. RRT had a find failure rate of 15% at the end of their used computation time, and A* found paths in case such exist. The exploratory analysis using deep reinforcement learning also demonstrated good performance but computing resources were demanding. System level validation showed reasonable real-world transfer of simulation results to the physical system, albeit with different absolute performance indicators due to sensor noise and wheel slip. The results give practical advice for choosing an algorithm depending on the application requirements, for instance on the complexity of the environment and computational resources available or desired level of reliability. The primary contribution of this paper is to make available empirical data that help practitioners select navigation algorithms suitable for various settings in mobile robot applications.Tétel Embargó alatt 3-phase BLDC motor analysis and control using Tensor Product TransformationAttia, Youssif Attia Aly; Taleb, Mayar; DE--Műszaki KarThis thesis designs and tests an angle-scheduled tensor-product (TP) state-feedback controller for speed control of a three-phase BLDC motor, modelled as an LPV system whose matrices depend on electrical rotor angle. The angle dependence is approximated by a 12-vertex polytopic model, and vertex gains are computed via LMIs with a common quadratic Lyapunov function; in real time the controller interpolates these gains and applies rate-limited, saturated phase voltages through a PWM interface. In nonlinear Simulink simulations of a 0–3000 rpm step, the TP and composite PI+TP controllers achieve 5–6 ms rise time, 10–15 ms settling time and 3–4% overshoot, whereas a tuned PI loop shows similar rise time but about 12% overshoot and ~45 ms settling. Steady-state results show that PI and PI+TP have essentially zero error and negligible ripple, while pure TP leaves a small bias and ~75 rpm peak-to-peak ripple, so the PI+TP architecture is identified as the best compromise between fast dynamics and smooth steady state.Tétel Embargó alatt Vision Based Automated optical inspection for real time quality controlNegm, Shahd; Taleb, Mayar; DE--Műszaki KarThe thesis presents a compact, low-cost, vision-based inspection and rejection cell for mixed SKU household chemical bottling lines operating at realistic conveyor speeds. A single fixed-view USB camera with controlled ring light illumination and a matte background feeds a modular perception pipeline that performs SKU aware localisation, cap height profiling, label and color verification, expiry date OCR and silhouette based deformation analysis. A lightweight YOLOv8 detector provides robust bottle level region proposals, while physics based geometric measurements and rule based checks are used for the individual quality tasks instead of an end to end black box network. Per bottle decisions are converted into a binary reject signal for a PLC controlled electropneumatic pusher, with timing budgets derived analytically from conveyor flight distance, image processing latency and actuator dynamics to guarantee correct removal of defective units. Experiments on representative bottle types show perfect cap sealing classification and competitive performance on the other tasks, including 97.0% accuracy for deformation detection, 90.9% for label conformity and 87.9% for expiry code recognition using only commodity imaging hardware and open source software. The work demonstrates a practically validated, retrofittable AOI solution that combines modern deep learning with interpretable geometric features and structured data logging, and is therefore well suited to Quality 4.0 upgrades in small and medium sized bottling plants.Tétel Korlátozottan hozzáférhető Optimal Control Design-based Grey Wolf Optimization Algorithm for Suspension System in Electric VehicleAbdurrahman, Faisal Haruna; Babangida, Aminu; DE--Műszaki KarThis research focuses on improving electric vehicle suspension systems using optimal control design based on Grey Wolf Optimization (GWO). It integrates a PID Controller, optimized by the GWO algorithm, with a passive suspension system to better handle road disturbances. The goal is to enhance ride comfort and maintain consistent road-wheel contact by reducing the impact of surface irregularities. Additionally, the study compares the performance of the GWO-PID system with a Model Predictive Contol (MPC) system. The comparison is evaluated using the Integral-Time weighted Absolute-Error (ITAE) as the fitness function.Tétel Korlátozottan hozzáférhető Kinematic simulation of a FANUC CRX 10iA/L collaborative robot in Simscape Multibody environment for motion analysisHumphery, Rosemary Chiemerie; Korsoveczki, Gyula; DE--Műszaki KarGiven the state of technology, the field of robotics is developing into where humans and robots share workspaces. As a result, human and robots have been able to work side by side due to this new emerging robotic system known as COBOT. This paper deals with the motion simulation of FANUC collaborative robot in Simscape Multibody for performance evaluation. By integrating the kinematic and dynamic parameters into the FANUC cobots I will be able to visualize how it behaves in the virtual environment. I will be using the FANUC CRX 10ia/l for the simscape multibody physics-based modelling, various trajectories and payload that are associated with each joint. This simulation done in MATLAB enables a detailed performance of the robot in terms of accuracy, repeatability and so on. The simulation results also shows valuable insights into the capabilities of the robot’s motion and reachability, which can be used for path planning, trajectory optimization, control system development, industrial applications, and educational purposes. With the use of this prototyping, we can use cobots in industries in a safe and effective way.