Department of Electrical Power Engineering and Mechatronics

Look in Estonian

Look in Estonian

JUHENDAJA: Johannes Muru

• Integration of a 6-DOF Collaborative Robot into a Flexible Manufacturing System
  The topic focuses on integrating the UR5 collaborative robot into a Festo Flexible Manufacturing System (FMS). The objective is to design and implement a robotic application that aligns with the FMS's requirements, allowing the UR5 robot to perform tasks within the system's workflow.
  The research involves developing communication between the UR5 robot and the FMS master PLC, ensuring smooth data exchange and task coordination. This integration will enhance the overall flexibility and efficiency of the manufacturing line. The project will require a thorough understanding of both robotic systems, and different industrial communication protocols, enabling the successful deployment of a fully integrated solution.
• Real-time Energy Monitoring for SCARA Robot
  This thesis focuses on the development of an embedded system solution for real-time monitoring of KUKA SCARA robot's energy consumption. The system will continuously track (and visualise) energy usage across different control modes and provide real-time feedback for analysis.
  Using the energy data, the study will enable comparative analysis of energy consumption under different control modes. Based on this feedback, techniques like parameter optimization can be applied to minimize energy usage while maintaining the required performance for various industrial applications.

JUHENDAJA: Even Sekhri

• Improved Safety System for Railways by Detecting Track Anomalies using a Stabilizing Platform
  The main idea would be to detect the anomalies on the railway tracks to improve safety. A special stabilizing gimble would need to be designed to mount the necessary sensors - Vision Sensor(s) and Radar Sensor(s). The complexity here would be to deal with ‘Moving Sensor and Moving Object’ for which robust systems are not available. The Radar System will work as a primary system in foggy and rainy days.
  The system should be able to warn the train driver about the anomalies on the track. The train should be running up to 50 km/h and the camera should be able to see the anomaly 500 m ahead of it. Stopping the train running at this speed is very much possible and safe (usually there is no need to apply emergency brake in such a case).
• Differentiating Screws mounted on a Microwave cavity filter and using UR5 robot to turn the tuning screws.
  The main idea would be to come up with a robust algorithm where the UR5 robot, holding the screwdriver in a designed manipulator, will reach the exact location of each tuning screws dynamically. The filter may be placed at any angle on the table (i.e. no holding fixture will be there on the table to hold the cavity filter) within the defined boundary and the vision system will detect the corners of the cavity filter and will calibrate the robot accordingly.
• Reinforcement Learning from Human Feedback for Processing Scattering Parameters.
  Introduced in 2024, Reinforcement Learning from Human Feedback (RLHF) is a concept which uses regret-based model of preference instead of using a reward-model used in conventional Reinforcement Learning (RL). The idea is to train an agent (an algorithm) to learn the optimum strategy to find the dips in the Scattering Parameter (S-parameter) data.
• Vision-based estimation of Gear Parameters.
  The idea is to come-up with a complete vision setup and use generic web-camera(s) to estimate the key gear parameters. The designed vision system must estimate the macro-geometry parameters of the gears. It should be able to estimate at least a few micro-geometry parameters as well.
• DetectApp - Determining the gun type and the shooting distance for a firing incident.
  The main idea is to come-up with a Machine Learning (ML) classifier which can detect the type of gun used in a firing incident and estimate the shooting distance just by looking at the marking(s) generated by the bullet i.e. – hole(s) and scattered gun powder on the clothing. The dataset for this task will be collected by shooting a dummy (mannequin) by at least 3 different guns in a controlled environment. The imaging setup will be installed, and the images will be clicked from a pre-defined distance. One of the main aspects to consider in this work would be to understand and consider the projectile motion of each bullet. Each bullet has its own trajectory even if it is shot at the same target from the same distance. This data is usually provided by the bullet makers and can be obtained from the internet. The goal would be to come up with an app for police/crime investigators to act and investigate faster in case of a firing incident just by scanning the clothing of the person who was shot from the recommended distance.
  Later (not under the scope of this thesis work), this work may be extended further to inclusion dataset of more types of bullets and making the app available to the police administration.

SUPERVISORS: Siarhei Autsou ja Anton Rassõlkin

• Diagnosis of mechanical parts of an industrial robot for eliminating undesirable consequences.
  In this project, student should make a mathematical model main mechanical parts of industrial robot and make modelling experiments. After that results of the modelling and real data should be compare.
• Develop control system with servodrive for eliminating undesirable oscillations.
  In this project, control system for servodrive with frequency convertor should developed. Control system should react on occurring oscillations and eliminate it.   

SUPERVISOR: Rolando Antonio Gilbert Zequera

• Battery Modeling for Direct Current (DC) converters in Electric Vehicle applications.
  Description: First, the student will learn about the different type of models for Battery Energy Storage Systems (BESSs). Second, the most optimal models for DC converters will be analyzed and implemented, studying their advantages and drawbacks. Finally, promising application in Electric Vehicles will be studied, taking into consideration the performance of different models.
  Recommended skills: It is recommended that the student has previous experience with coding or willing to learn new programming languages (Python, R, MATLAB), interest in energy storage systems, and good background in calculus.
  Learning outcomes: At the end of the thesis, the student will understand the implementation of Battery Energy Storage Systems in different study cases, considering Health and Charge indicators, operating data, and degradation mechanisms.

SUPERVISOR: Mahmoud Ibrahim

• Development of a Magnetic Transmission unit for an EV Motor.
  The main research outcome is to develop a prototype of a magnetic transmission unit for an EV motor. The magnetic transmission unit refers to a magnetic gearbox that resembles the traditional mechanical gears in geometry and function, using magnets instead of teeth.
  Theoretically: Students will study the concept of different magnetic gearing modes. The student will prepare a comparative study between magnetic and mechanical gears. The student has to build a virtual model of the developed transmission unit.
  Practically: The student must build a physical prototype of the developed transmission unit.
• Virtual-Physical Integration and Fault Detection in Electric Motor for Digital Twin Development.
  Description: This thesis focuses on the integration of virtual and physical twins of the electric motor to establish a reliable digital twin. The aim is to develop a communication framework between the virtual and physical domains, enabling real-time data exchange and synchronization. Furthermore, fault detection algorithms will be implemented to monitor and diagnose motor malfunctions. Tools like ROS (Robot Operating System), MATLAB/Simulink, and hardware interfaces (sensors, data acquisition devices) will be utilized.
  Expected Outcomes:
  Development of a communication interface between the virtual and physical twins, enabling bidirectional data exchange and synchronization.
  Implementation of fault detection algorithms to identify abnormalities, such as motor overheating, bearing failures, or rotor imbalances.
  Real-time monitoring and visualization of motor health status through the digital twin interface.
  Comparative analysis of simulated and physical motor behavior to evaluate the accuracy and reliability of the digital twin.
  Importance: This thesis addresses the crucial aspect of virtual-physical integration and fault detection in the context of a digital twin for the electric motor. The developed framework can facilitate predictive maintenance, enhance motor reliability, and enable efficient utilization of the autonomous electric vehicle's propulsion system.
• Real-Time Communication and Data Synchronization between Virtual and Physical Models for Digital Twin Integration.
  Description: This thesis aims to develop a robust communication framework that enables real-time data exchange and synchronization between the virtual and physical models of the electric propulsion system. The focus will be on establishing bidirectional communication channels, implementing data fusion algorithms, and addressing latency issues. Tools such as OPC (OLE for Process Control), MQTT (Message Queuing Telemetry Transport), and network simulation software can be utilized.
  Expected Outcomes:
  Design and implementation of a communication framework that enables real-time data exchange and synchronization between the virtual and physical models.
  Development of data fusion algorithms to combine information from both domains and generate a unified representation of the electric propulsion system.
  Evaluation of the communication framework's performance in terms of latency, reliability, and data consistency.
  Comparative analysis of the integrated digital twin's behavior against real-world scenarios to assess the accuracy and effectiveness of the communication approach.
  Importance: This thesis addresses the crucial aspect of establishing seamless communication between virtual and physical models. The developed communication framework will facilitate accurate representation and analysis of the electric propulsion system, enabling effective decision-making and control in an autonomous electric vehicle.

SUPERVISOR: Martin Sarap

• Additively Manufactured Solutions with Optimized Topologies for Electrical Machine Thermal Applications.
  The aim of the work is to create an additively manufactured (3D printed) cooling solution for an electrical machine by utilizing topology optimization. To achieve this, the student will create a 3D geometry, model its effectiveness using simulation software (Comsol Multiphysics) and then optimize the geometry until the maximal cooling performance with the given restrictions is achieved. Finally, a prototype of the model is created through additive manufacturing for real-world testing.
  Theoretical part: thermal calculations, computational fluid dynamics, additive manufacturing, cooling of electrical machines, modelling, topology optimization.
  Practical part: manufacturing the prototype from metal, performing the measurements.

SUPERVISORS: Indrek Roasto / Dmitri Vinnikov

• Power electronics systems for residential microgrids

SUPERVISOR: Oleksandr Husev

• Research and Development of the Artificial Intelligent Based Control System for Power Electronics Applications.
• Research and Development of GaN Based Power Electronics.
• Wireless power transfer (scooter contactless charging).

SUPERVISOR: Andrii Chub

• Research and Development of High Step-up dc-dc Converters.
• Partial Power dc-dc Converters.