Robotics and Computer Engineering - Master's theses

Permanent URI for this collection


Recent Submissions

Now showing 1 - 20 of 95
  • Item
    ESTCube-2 Mission Control System: Preparation for In-orbit Operation
    (Tartu Ülikool, 2023) Toomast, Cathy; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Space missions rely on the mission control system (MCS) for spacecraft health monitoring, commissioning, routine operations, and emergency procedures. However, often the MCS stays in the shadow of the satellite itself, creating a situation where the satellite is launch-ready, but the system necessary to operate the satellite from the ground is not. This is also the case with ESTCube-2, a satellite that started development in 2016 and is soon ready for launch. ESTCube- 2 is a spacecraft mainly developed by student volunteers, and its main mission is to demonstrate deorbiting with plasma brake technology [1]. To operate the spacecraft, there is a need for a functional mission control system. The MCS has been developed through the years, but when it was initially built, the satellite was not ready to be tested with the system. For this reason, the system is not yet widely used and has unresolved issues. The author of the thesis investigates and lists the actions that need to be taken to have an operational MCS by the start of the mission. Furthermore, to understand the needs of mission operating systems, the author used a qualitative research method, interviewing four people with experience with operating satellites. For the final system to be useful for the mission, the author made changes in the mission control system during the thesis and implemented suggestions from interviews with spacecraft operators. Initial tests on the ground were performed with the ESTCube-2 engineering model. Additionally, the author will list ideas and notes regarding what could be done better in the future at the end of the thesis.
  • Item
    XR Teleoperation Demo Development
    (Tartu Ülikool, 2023) Zorec, Matevž Borjan; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    This thesis designs an educational real-time visual feedback teleoperation demonstration. The importance of a good user experience is highlighted while showcasing the feasibility of using open-source solutions such as Godot Engine version 4 for teleoperation setups. Reviewed literature narrowed design requirements, outlining that a representative teleoperation demonstration could provide a positive experience, intuitive movement control, direct real-time visual feedback for teleoperation and be open-sourced, with user and video stream evaluations as research objectives. Employing design thinking, 'RoverXR' is iteratively developed with M5 RoverC-Pro for movement and serving WebSocket protocol real-time Motion JPEG high-definition video from Raspberry Pi v2.1 Camera Module via a Raspberry Pi Zero. Custom MPV player and Godot scenes were prepared, featuring video stream playback and providing a virtual user interface on the Meta Quest 2 headset. User evaluation participants report a positive, engaging experience and provide helpful feedback, showcasing the potential of low-latency, high-quality video streaming, and virtual scene representation in teleoperation demonstrations for educational purposes.
  • Item
    Sampling-based Bi-level Optimization aided by Behaviour Cloning for Autonomous Driving
    (Tartu Ülikool, 2023) Shrestha, Jatan; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Autonomous driving has a natural bi-level structure. The upper behavioural layer aims to provide appropriate lane change, speeding up, and braking decisions to optimize a given driving task. The upper layer can only indirectly influence the driving efficiency through the lower-level trajectory planner, which takes in the behavioural inputs to produce motion commands for the controller. Existing sampling-based approaches do not fully exploit the strong coupling between the behavioural and planning layer. On the other hand, Reinforcement Learning (RL) can learn a behavioural layer while incorporating feedback from the lower-level planner. However, purely data-driven approaches often fail regarding safety metrics in dense and rash traffic environments. This thesis presents a novel alternative; a parameterized bi-level optimization that jointly computes the optimal behavioural decisions and the resulting downstream trajectory. The proposed approach runs in real-time using a custom Graphics Processing Unit (GPU)-accelerated batch optimizer and a Conditional Variational Autoencoder (CVAE) learnt warm-start strategy and extensive experiments on challenging traffic scenarios show that it outperforms state-of-the-art Model Predictive Control (MPC) and RL approaches regarding collision rate while being competitive in driving efficiency.
  • Item
    Comparison of Water Detection Models for an Off-road Unmanned Ground Vehicle
    (Tartu Ülikool, 2023) Rustambayli, Fidan; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Water hazards can cause unmanned ground vehicles (UGVs) to become stuck or break down during an autonomous mission, damage electronic components and sensors, and require costly repairs or replacements, making it crucial for UGVs to identify water hazards in real-time, determine secure path around them, or reduce their speed when appropriate to cross them safely. This thesis proposes a water detection system for UGVs in off-road environment. The proposed approach combines convolutional neural networks (CNNs) with transfer learning, leveraging their capabilities for effective water detection. The thesis includes a comprehensive review of traditional sensor-based methods and recent deep learning-based techniques. Real-world data collected in off-road environments are utilized to evaluate the proposed approach, and the method achieves a 0.50 Mean-IoU score and 92.74% accuracy on the test dataset. We also include a comparative analysis of the method with a previous deep learning-based semantic segmentation method for water detection. The comparison provides insights into the relative strengths and weaknesses of these approaches for water detection in off-road environments. Overall, this thesis provides valuable insights into the use of deep learning for semantic segmentation in challenging environments.
  • Item
    LEAN metoodika rakendamine aktsiaselts Chemi-Pharm tootmises
    (Tartu Ülikool, 2023) Nigul, Mihkel Erich; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    LEAN metoodika baseerub Toyota tootmissüsteemil, mis loodi 20. sajandil Jaapani töötusinsener ja ettevõtja Taiichi Ohno poolt. Taiichi Ohno poolt loodud süsteem oli revolutsioon masstootmises, sest see võimaldas tõsta märkimisväärselt tootmise efektiivsust ja vähendada praaktoodete arvu. Toyota tootmissüsteem on tänaseni ajakohane, mistõttu on ka käesoleva lõputöö eesmärk Taiichi Ohno meetodite rakendamine kaasaegses tootmisettevõttes. Käesolevas lõputöös uuriti ja rakendati LEAN metoodika tööpõhimõtteid ja automatiseeriti ebaefektiivsed tootmisprotsessid. Lõputöö praktilises osas teostati villimisliinile parendustöid, kasutades PDCA probleemi lahendamise tsüklit. Kõigepealt fikseeriti algseis, uuriti villimisliini tööd ja kaardistati kitsaskohad, leiti probleemide juurpõhjused ja pakuti välja võimalikud lahendused. Seejärel viidi ettepanekud ellu, analüüsiti parendustööde kasu ettevõttele ja standardiseeriti lõputöö käigus lõputöö praktilises osas arendatud parendused.
  • Item
    Leveraging neural models for data processing and analysis automation
    (Tartu Ülikool, 2023) Kõiv, Erik; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Unmanned Ground Vehicles (UGVs) are a staple in some industries and are entering the market in others. Development of these UGVs and their automation is resource intensive and timeconsuming work. Specifically the job of processing and analysing data collected by the various sensors and cameras has so far been done by human workers. In recent years however, it has become possible to propose the automation of these tasks. This thesis describes the development of a pipeline application aimed at reducing the workload of the workers doing these jobs by leveraging neural models such as CLIPSeg, capable of zero-shot text-prompt image segmentation, to extract data from video frames based on specified classes of interest. A proof of concept demo was developed and presented to potential users, leading to the extraction of requirements for a minimum viable product (MVP). The MVP requirements included avoiding image resizing distortion, a command-line interface, and additional post-inference data analysis. The CLIPSeg model was evaluated alongside CLIPSurgery, another zero-shot image segmentation model, using a testing dataset. CLIPSeg demonstrated higher viability for the selected classes and was further evaluated using an 80% model score and 0.05% image area threshold to eliminate false positive results with great success. The final MVP application fulfilled all presented requirements and proved the viability of the CLIPSeg model for the use-case
  • Item
    Software development for a high-speed digitizer to provide online access to longitudinal bunch profiles in the Large Hadron Collider
    (Tartu Ülikool, 2023) Jõul, Jüri; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Particle accelerators such as the Large Hadron Collider (LHC) strive for high luminosity in order to produce as much data for physics experiments as possible. Measuring the performance of an accelerator is a crucial step in improving it. Accelerator physicists have proposed novel diagnostic methods utilising longitudinal bunch profiles from wall current monitors. So far at CERN, online access to these profiles with the required performance to enable new diagnostic methods did not exist. As a result of this thesis, a proof-of-concept software solution to acquire and provide online access to wall current monitor data from the LHC was created. In the process, the limitations stemming from the existing hardware and software resources, as well as the users’ needs were considered. The software solution along with its component algorithms was validated with the acquisition of real wall current signals from the LHC.
  • Item
    An Open-Source Robotic Study Companion for University Students
    (Tartu Ülikool, 2023) Baksh, Farnaz; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    This thesis presents an evidence-based approach and develops an affordable and effective Robotic Study Companion (RSC) prototype for university students. It addresses the lack of social robots tailored to higher education and the scarcity of open-source educational platforms. Through a comprehensive literature review on Human-Robot Interaction (HRI), social and companion robots, and Natural Language Processing (NLP) technologies, this work identifies trends and best practices for educational social robots. The research systematically reviews select social robots, examining their applications, technical features, design, and human-centric interaction. It explores the human perspective of HRI, focusing on users in educational settings. Based on these insights, functional and non-functional requirements are established for the RSC, inspiring its design and development. The RSC prototype, built using off-the-shef components and leveraging OpenAI's large language models, demonstrates its potential to simplify complex concepts for students. The long-term goal is to enhance the RSC's design, durability, and commercial viability.
  • Item
    Evaluation and Optimization of Feature Detectors Towards Off-Road Visual Odometry
    (Tartu Ülikool, 2023) Tepandi, Tarvi; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Feature detectors are used in many computer vision applications, including Visual Odometry (VO). However, their utilization in off-road VO remains a topic of great interest. In this body of work, software tools are developed in order to evaluate and parameter-optimize feature detectors towards real-time off-road VO. Using the tools developed, various classical as well as state-of-the-art machine learning-based feature detectors are evaluated and optimized, including their usage in a pre-existing VO implementation and analyzing the output trajectory. The analysis and results presented show that although the quality of feature detectors have an impact, optimizing them alone cannot overcome the inherent drawbacks of monocular VO approaches. Based on the analysis, recommendations of potential feature detectors that can support real-time off-road VO are made and optimized parameters of selected feature detectors are provided. Key areas of improvement for future research in the field are also identified.
  • Item
    Assessing Event-Based Localization Algorithms for Vehicular Off-Road Applications
    (Tartu Ülikool, 2023) Salumets, Sten; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Event cameras, with their high temporal resolution and dynamic range, represent a promising technology for localization applications. Yet, their performance in off-road environments remains untested. This thesis addresses this gap by assessing three eventbased localization methods in off-road settings. A conventional frame-based method is included as a benchmark for comparison. The effectiveness of each method is assessed by comparing computed trajectories with ground truth data. The findings indicate that current publicly available event-based methods are not yet mature enough to provide accurate and robust performance in off-road environments. The study underscores the need for further research to fully harness the potential of event cameras in these challenging settings.
  • Item
    Capturing the sound of reed instruments via measurement of reed strain
    (Tartu Ülikool, 2023) Pihlap, Meelis; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    The sound from reed instruments (such as the bagpipe or the clarinet) is conventionally captured using microphones. Microphones capture all sound regardless of source, making it a volatile choice in live settings, where background noise (e.g. from other instruments) is prominent. As such, an instrument pickup for reed instruments is desirable for use in settings where an isolated signal is required. There are few existing reed instrument pickups and their installation often involves permanent changes to the instrument. A novel reed instrument pickup is introduced, which uses a strain gauge mounted directly to the oscillating tongue as the sensing element. The pickup is compared against a measurement microphone and the results show that after a filtering stage, the pickup sounds good enough to compete with the microphone.
  • Item
    Reverse Engineering a Car Immobilizer
    (Tartu Ülikool, 2023) Laks, Jürgen; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Car immobilizers are devices that prevent car thefts by immobilizing cars if a specific token, usually carried by the intended driver, is not in proximity to the car. This work documents the steps taken to reverse engineer the Skybrake DD2+ immobilizer in order to discover its security weaknesses. These steps include capturing the data that is transferred between the components of the immobilizer and analyzing the captured data. The firmware image of the immobilizer was dumped and analyzed. This work presents multiple possible attack vectors that this immobilizer is vulnerable to. Most notably this work demonstrates methods for bypassing the layer of security the immobilizer is supposed to provide.
  • Item
    Vedelikupõhine mälu kiukoodis
    (Tartu Ülikool, 2023) Küüts, Mona; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Kogu inimkonna ajaloo vältel on tekstiilidel alati olnud märkimisväärne roll meie elus, alustades tekstiilist kui elementaarsest ellujäämise vahendist, lõpetades globaalse eneseväljenduse vahendina moetööstuses. Lisaks eelnevalt mainitule paigutub sellele spektrile ka palju funktsionaalseid tekstiili rakendusi, millest üks on tekstiilidega integreeritud kantavad robotid. Pehmest tekstiilist ajamid on pehmerobootika valdkonnas laialt levinud tänu oma paindlikkusele, kergusele ja ohutusele. Lisaks sellele, et tekstiil käitub täiturmehhanismide sees tugistruktuurina, on tekstiilil ka suurepärased kapillaarsusomadused mis pole robootikas leidnud laialdaselt rakendust. Lõputöös demonstreeritakse kuidas tekstiili niitide programmiline paigutus (kiukood) aktiivse poorse materjali sees võimaldab määrata vedeliku (lahusti ja elektrolüüt) eelistatud liikumise suuna. Lahusti liikumine mööda tekstiili niitide kapillaarpindu võimaldab elektrolüüdi hulga lokaalseid suurenemisi või vähenemisi, mis omakorda võimaldab muuta tekstiili omadusi. Selle uurimiseks tikiti tekstiil (kiukood), mis osaliselt kaeti elektroaktiivsete materjalidega. Elektroaktiivsete materjalidega kaetud tekstiil moodustas aktiivse ala. Antud töö raames kasutati lahustina 4-metüül-2-pentanooni, et transportida elektrolüüti (ioonset vedelikku) tekstiili passiivsete ja aktiivsete alade vahel. Lahusti aurustumise järel on võimalik elektrolüüdi kontsentratsioonigradientidesse salvestada informatsiooni (vedelikupõhine mälu). Pehmed aktiivsed tekstiilid on väga nõutud tervishoiuvaldkonnas, kus on suur vajadus eksoskelettide järele, mida saaks kanda nagu rõivaesemeid, mis ühilduks inimese kehaga ning mille materjali omadusi on võimalik muuta ja kontrollida.
  • Item
    Augmented reality (AR) for enabling human-robot collaboration with ROS robots
    (Tartu Ülikool, 2022) Rybalskii, Igor; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    With current industrialization trends on the reintroduction of human into the manufacturing process and development of augmented reality I propose the interface, which uses Augmented reality to allow the operator to interact with robotic systems, such as manipulators and mobile robots. 2 interfaces were created: one for manipulators and one for mobile robots. Both of them are developed to work on Microsoft Hololens 2 augmented reality glasses and robots, running ROS. Interface for manipulators allows user to control the robot by sending end-effector goals and previewing the goal joint states of the manipulator. Interface for mobile robots allows the user to send navigation goals and preview the robot’s movement trajectory. Interfaces were developed using Unity game engine. Developed interfaces were tested with UR5e manipulator and Robotont mobile robot.
  • Item
    Sending and storing data from a smart scoliosis brace
    (Tartu Ülikool, 2022) Metsala, Nathan; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Scoliosis is a widely known disease that troubles many people globally. Scoliosis is treated with a specialized brace and assigned therapy, which consists of exercise routines that are recorded in patient journals. Often the exercise routines incorporate the medical brace. Outside of patient trust and therapy results, there is no universally agreed upon method of monitoring brace usage and therapy exercise routines. The aim of this thesis is to provide an example of how modern technology can be used to improve scoliosis therapy and help both patients and doctors see better results when treating scoliosis. The thesis aims to provide more scientific data for doctors to more effectively treat the disease.
  • Item
    Object search and retrieval in indoor environment using a Mobile Manipulator
    (Tartu Ülikool, 2022) Maniruzzaman, Md.; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    Robots are increasingly viewed as service agents in offices and homes. In many countries where the average population is aging, robots can be used for elderly care. This Thesis explores one such possibility using a mobile manipulator robot. Such robots have a mobile base to move from one place to another and an arm to pick and place objects. This Thesis considers a problem where the mobile manipulator needs to search for an object in an environment and bring it to some location. The optimal object search is formulated in terms of the popular traveling salesman problem (TSP) that computes the optimal sequence in which the Robot can visit all the possible locations where the object can possibly be. Prior information about the more likely locations is brought in by scaling the edge-weight of the TSP graph through the probabilities of the location. The Thesis can combine the output of TSP with navigation and manipulation planning built on top of Robot Operating Systems (ROS) to build the complete object search and retrieval pipeline. The results of the Thesis are validated both in simulation and actual hardware experiments.
  • Item
    Assessment of ethnic and gender bias in automated first impression analysis
    (Tartu Ülikool, 2022) Krull, Friedrich; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    This thesis aims to investigate possible gender and ethnic biases in state-of-the-art deep learning methods in first impression analysis. Analysing a person with some software, businesses want to find the best candidate, without the person being judged by their gender or ethnicity. To achieve this, a first impression dataset about the big five personality traits, with additional information about the person’s gender and ethnic background, was used. Biases were both investigated with models trained on balanced and imbalanced data, where balanced here refers to the number of frames used from people classified as Asian, African-American, or Caucasian in the dataset. The results with both the balanced and imbalanced datasets were similar. With all the models the accuracy for Asians was much higher compared to others, which may come from the fact that the dataset did not include enough variance in the Asian data, so when evaluating, all Asians were seen similarly.
  • Item
    Design, Testing and Analysis of an Affordable Direction of Arrival System Using Off-The-Shelf SDR Hardware and Software
    (Tartu Ülikool, 2022) Amor, Erik; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    The growing use of wireless communication increases the need for devices to determine the location of the signal source. Affordable, accessible and universal devices can have a great role in addressing this aside precision tools. Combining popular off-the-shelf radio receivers with direction of arrival algorithms could provide a convenient tool. The setup is proven to give sufficient results for several location applications using HackRF radio and signal processing in software. The accuracy of the system can be improved by characterising the system for each use case. After the setup is completed, such system can still be convenient to use for non-technical end user.
  • Item
    Improving object detection in adverse weather conditions for Auve Tech’s autonomous vehicle
    (Tartu Ülikool, 2022) Abbasov, Elmar; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    As autonomous vehicles are becoming more prominent in our lives we want their computing systems to be able to recognize objects with the best accuracy possible, regardless of the weather conditions. In order to achieve better accuracy with machine learning based visual object detection we compare 2 approaches: training an object detection neural network with synthetic rain added images and removing rain from images using a different state-of-the-art neural network before feeding them to an object detection neural network trained with non-rainy images.
  • Item
    Synthetic Vascular Tissue with Binary Control for Dynamic Stiffness
    (Tartu Ülikool, 2022) Ratas, Herman Klas; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
    In the scope of this work, a synthetic vascular tissue with binary control for dynamic stiffness was built. The task had two distinct objectives – design and build a vascular cellular system for dynamic stiffness and design and build a soft distributed valve. The built cellular system and soft distributed valve were characterized and showed dynamic stiffness and state control. These properties give them great potential for integrating them into midwifery simulation training for more seamless state control