Education > Theses > Bachelor

2025

1. Device for generating prescription maps using embedded AI in precision agriculture

   Estela Mora Barba

   Supervisor: Néstor Lucas-Martínez

   Sep 2025

   Abs Bib PDF Website

   The objective of this project, entitled “Device for generating prescription maps using embedded AI in precision agriculture” is the design and implementation of an autonomous hardware system based on an Unmanned Aerial Vehicle (UAV). This system is capable of capturing agricultural images, processing them locally using artificial intelligence (AI) models, and generating prescription maps in real time, without relying on external infrastructures. The proposed solution aims to increase the autonomy and responsiveness of agricultural operations, improving efficiency and sustainability through the use of edge computing, modular service-oriented architecture, and open-source technologies. The system builds upon concepts and tools from European projects such as AFarCloud and FlexiGroBots, as well as integration with the Fleet Manager platform for mission planning and supervision. This enables the dynamic coordination of multiple UAVs and optimises resource allocation in the field. From a technological perspective, the UAV integrates RGB cameras, an embedded computing module with an AI hardware accelerator (Hailo-8 and Hailo-8L), and ROS2 middleware for internal communication and mission control. The architecture relies on gRPC for inter-service communication, REST for interaction with external services, and ThingsBoard for real-time telemetry and results visualisation. In addition, Docker containers are employed to ensure portability and facilitate deployment across different platforms. Regarding economic constraints, low-cost hardware and open-source software were prioritised, reducing expenses and favouring system replicability. From an environmental standpoint, the system’s precision in treatment application and early detection of diseases contributes to a more efficient use of resources. This approach reduces environmental impact and promotes sustainable agricultural practices aligned with the Sustainable Development Goals (SDG). The development methodology was carried out in two phases. The first, design, involved defining the architecture, information models, UML diagrams, and the selection of optimal technologies. The second, implementation, included the development of microservices, technology integration, creation of workflows for image capture, analysis, and prescription map generation, and validation in simulated environments under different scenarios. Validation, conducted in accordance with the ISO/IEC/IEEE 29119:2020 standard, included functional and non-functional tests to verify the system. Embedded processing performance, communication robustness, and the system’s ability to operate continuously under variable conditions were also evaluated. The results show that the system is stable, efficient, and suitable for deployment in real agricultural environments. Future work includes expanding the platform to autonomous ground vehicles (UGV). Other enhancements involve improving AI models to cover more crop types and diseases, and incorporating multispectral analysis for more accurate diagnostics. Additionally, optimising communication protocols for connectivity-limited environments. These efforts aim to consolidate a comprehensive platform for intelligent fleet management in precision agriculture.

   @phdthesis{mora_barba_device_2025,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Device for generating prescription maps using embedded {AI} in precision agriculture},
     url = {https://oa.upm.es/90982/},
     language = {English},
     school = {Universidad Politécnica de Madrid},
     author = {Mora Barba, Estela},
     month = sep,
     year = {2025},
     keywords = {pdf: https://oa.upm.es/90982/1/TFG\_MORA\_BARBA\_ESTELA.pdf, sdg-12, sdg-13, sdg-2, sdg-9, supervisor: Lucas-Martínez, Néstor},
     supervisor = {Lucas-Martínez, Néstor},
     dimensions = {true}
   }

2. Conciencia situacional en tiempo real para los servicios de emergencia

   Arturo López López

   Supervisor: Néstor Lucas-Martínez

   Jul 2025

   Abs Bib

   The implementation of technological advances in emergency management can play a key role in performing the various tasks of emergency services, these advances that are already implemented in other areas contribute to a significant improvement in time and efficiency. This project seeks to improve the response capacity of the emergency services, making it faster and more efficient through a system that uses autonomous learning techniques. The objective is to collect and process data from various sources in the environment of the city of Madrid. The project is structured in two main parts, in the first one a series of prediction models are trained and evaluated with a set of transformed and enriched data to obtain better metrics. The second part seeks to use this model to generate an alert system that informs about the zones and time slots with the highest probability of accidents using a messaging application. The architecture of the system has been designed so that the model can be tested, and weather data can be simulated daily, using free software and intuitive and accessible development environments. At the functional level, tests have been validated such as the assignment of districts to their weather stations, the generation of alerts above the decision threshold and the sending of these alerts to the application. In the non-functional section, parameters such as execution time, system robustness and message usability have been evaluated. This project has a wide impact in social, public health, economic and technological fields, being able to take preventive measures against risk areas and thus reduce the accident rate and improve road safety, lower direct costs of those produced by accidents, such as property damage or health care, and indirect costs, such as sick leave and an advance towards more efficient and safer smart cities through technology. Although the results obtained are not optimal as it addresses a complex issue with multiple factors, which derives in a significant technical challenge, through a more specific approach, a larger data source and numerous tests for the prediction model and, together with the alert system associated with a weather data API, it could provide an effective tool for emergency services.

   @phdthesis{lopez_lopez_conciencia_2025,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Conciencia situacional en tiempo real para los servicios de emergencia},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {López López, Arturo},
     month = jul,
     year = {2025},
     keywords = {supervisor: Lucas-Martínez, Néstor},
     supervisor = {Lucas-Martínez, Néstor},
     dimensions = {true}
   }

3. Artificial intelligence applied to cybersecurity in ICT networks

   Ainara Picot Arizmendi

   Supervisor: Néstor Lucas-Martínez

   Jul 2025

   Abs Bib

   This project lies at the intersection of cybersecurity and autonomous aerial systems. Its main goal was to design and implement an intelligent, real-time detection system for cyberattacks targeting UAV telemetry using hybrid ML techniques. The proposed solution addresses the urgent need to secure wireless communication channels in unmanned vehicles, which are vulnerable to Wi-Fi de-authentication or command injection attacks. Various technological and environmental constraints shaped the design: the system had to be lightweight and self-contained, capable of running offline on embedded devices, and optimized for real-time processing without high latency. Economically, it was built to be modular and affordable by leveraging open-source tools and reusable components. The methodology combined multiple AI techniques applied to the ECU-IoFT dataset. After preprocessing and manual balancing, categorical features were encoded and numerical fields normalized. Anomaly detection was conducted via CBLOF, followed by feature compression with an autoencoder and supervised classification using XGBoost. To improve the multiclass model’s generalization, controlled noise was injected into the majority class labels. This strategy increased robustness and mitigated overfitting. The system underwent comprehensive functional and real-world testing using tools such as aireplay-ng and tcpdump. Results were highly promising: the binary model achieved an F1-score exceeding 95%, while the multiclass model, after label noise injection, accurately identified all four traffic types (normal, de-authentication, WPA2 cracking, and API misuse). In summary, this project has produced an effective, adaptable, and deployable solution with potential applications in autonomous surveillance or cyber-defense scenarios. It adheres to principles of sustainability, computational efficiency, and ethical AI usage, and lays the foundation for future extensions, including integration with real drone platforms and expanded attack coverage.

   @phdthesis{picot_arizmendi_artificial_2025,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Artificial intelligence applied to cybersecurity in {ICT} networks},
     language = {English},
     school = {Universidad Politécnica de Madrid},
     author = {Picot Arizmendi, Ainara},
     month = jul,
     year = {2025},
     keywords = {supervisor: Lucas-Martínez, Néstor},
     supervisor = {Lucas-Martínez, Néstor},
     dimensions = {true}
   }

4. Sistema de diagnóstico precoz de enfermedades cutáneas

   Javier Bravo Acedo

   Supervisor: Néstor Lucas-Martínez

   Jul 2025

   Abs Bib

   The objective of this final degree project is to develop an automated early dermatological diagnosis system, designed to be integrated and deployed in any location with internet access, using a distributed client-server architecture. This system, based on Yolo object detection technology, enables real-time detection of skin conditions, significantly improving access to medical care, reducing dermatologist’s workload, and shortening response times in the early care stages of the chosen healthcare system for its implementation. The system‘s design is based on the urgent and verifiable need to offer an efficient, accessible, and reliable solution in a context where the lack of early dermatological care has been worsened by environmental factors such as the depletion of the ozone layer and increased UV radiation. These factors have contributed to a rise in skin disease cases globally, clearly highlighting the need for a system that enables early, accurate diagnosis with nearby and professional access, addressing this international challenge. The design decisions of the proposed system are based on the need to facilitate fast and cost-effective diagnoses. For this reason, Yolo was chosen due to its ability to process images in real time with high object detection accuracy, making it an ideal and robust tool for skin condition detection. Additionally, a modular and user-friendly design was chosen to ensure the system‘s accessibility, scalability, and ease of maintenance. Convolutional Neural Networks were selected for their ability to analyze spatial data and computational efficiency, ensuring accurate detection of various types of skin conditions under different lighting conditions and skin characteristics, regardless of user positioning or lighting environment. To improve model accuracy, the Transfer Learning technique was implemented, leveraging pre-trained models such as Yolov8, chosen for its balance between accuracy and speed. This approach allows for a more efficient development process with lower computational resource consumption and easier implementation compared to other models. Furthermore, the use of datasets collected from various research projects, as reflected in the references, includes a wide variety of annotated dermatological images, which is crucial for training the model and ensuring its generalization capability in identifying different skin conditions. This proof of concept uses a system based on Machine Learning, specifically supervised learning, to achieve efficient classification of skin diseases. The system includes an intuitive user interface, consisting of a series of menus that guide the user through the diagnostic process. This interface is designed for user identification through a national identity document, stored in a database along with user main data and report history, which would facilitate integration with healthcare systems that comply with international standards.

   @phdthesis{bravo_acedo_sistema_2025,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Sistema de diagnóstico precoz de enfermedades cutáneas},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {Bravo Acedo, Javier},
     month = jul,
     year = {2025},
     keywords = {supervisor: Lucas-Martínez, Néstor},
     supervisor = {Lucas-Martínez, Néstor},
     dimensions = {true}
   }

5. Smart management system for unmanned autonomous vehicles in precision agriculture on edge

   Estela Mora Barba

   Supervisor: Néstor Lucas-Martínez

   May 2025

   Abs Bib PDF Website

   The objective of this project, titled “Smart management system for unmanned autonomous vehicles in precision agriculture on edge”, is the design and implementation of an intelligent fleet management system based on edge architecture for Unmanned Aerial Vehicles (UAVs) used in precision agriculture tasks. The proposed solution aims to improve the autonomy, efficiency, and sustainability of automated agricultural operations through the use of advanced software technologies, distributed architecture, and microservices. It also draws on information models and workflows derived from the European AFarCloud project, which is focused on the evolution, automation, and optimisation of the agricultural sector. The development has focused on optimising the coordination and autonomous execution of agricultural missions. This is achieved by implementing a system capable of managing task dependencies, dynamically updating the status of UAVs and offering visualisation platforms and tools for continuous monitoring and analysis. From a technological point of view, the system is built on a microservices architecture, using ROS2 as middleware for communication with UAVs and gRPC as the internal protocol between microservices. REST is used for the interaction with external web services, while MariaDB serves as the database management system. For real-time visualisation of telemetry and device status, the ThingsBoard platform is integrated. With respect to economic constraints, the project has deliberately opted for open-source software and tools. This has helped reduce development and implementation costs while facilitating system replication and scalability. From an environmental perspective, the system promotes more sustainable agriculture. Improves the efficient use of natural resources, such as water, fertilisers, and pesticides, thanks to the precise and automated execution of tasks via UAVs. The development methodology is divided into two phases. Design and implementation. In the design phase, information models, system architecture and UML diagrams were defined. The implementation phase involved the development of various microservices, the integration of technologies, and the simulation of the behaviour of the system in controlled testing environments. The system was validated through functional and non-functional tests based on the ISO/IEC/IEEE 29119:2020 standard. These tests verified the correct execution of the mission, communication between components, system latency, and data consistency. The results demonstrate that the system is stable, efficient, and viable for deployment in real-world scenarios. In future work, it is proposed to improve decision-making algorithms, integrate Unmanned Ground Vehicles (UGVs) and optimise communication protocols, with the aim of consolidating a complete platform for intelligent fleet management in precision agriculture.

   @phdthesis{mora_barba_smart_2025,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Smart management system for unmanned autonomous vehicles in precision agriculture on edge},
     url = {https://oa.upm.es/89100/},
     language = {English},
     school = {Universidad Politécnica de Madrid},
     author = {Mora Barba, Estela},
     month = may,
     year = {2025},
     keywords = {keywords:, pdf: https://oa.upm.es/89100/3/TFG\_MORA\_BARBA\_ESTELA.pdf, sdg-15, sdg-2, sdg-8, sdg-9, supervisor: Lucas-Martínez, Néstor},
     supervisor = {Lucas-Martínez, Néstor},
     dimensions = {true}
   }

2024

1. Diseño de una aplicación móvil para la detección de smishing dirigida a personas mayores

   Rubén Sanz Barrio

   Supervisor: Néstor Lucas-Martínez

   Dec 2024

   Abs Bib PDF Website

   This paper presents the project brief entitled "Design of a mobile application for smishing detection aimed at elderly people". In recent years there has been an increase in phishing cases, specifically 534% between 2018 and 2023 according to APWG reports. Phishing also accounts for 43% of digital crimes in 2023, according to reports from the FBI’s Internet Crime Complaint Center, representing almost twice as many cases as the second-most-committed cybercrime. In this context, the elderly people are the most affected age group, with losses of almost $3,500,000,000 in 2023. Given all the above, the overall purpose of the project is to develop a mobile application that enables older people to detect phishing, specifically smishing, in a simple and accessible way. First, the task of classifying SMS is carried out by a Machine Learning model based on the characteristics of URLs extracted from the content of messages. This required an analysis of the different results obtained by the models studied. Secondly, the data models to be stored in the database were designed to ensure that their structure would meet all the proposed functionalities. Finally, the development of the mobile application was divided into frontend and backend assigning responsibilities according to the MVC architectural pattern. This separation allowed the user interface to handle representation and interaction, while the backend handled data handling and business logic. Since the mobile app is aimed at older people, there are several factors that were considered. To access the application, it is not necessary to use emails, usernames, passwords, etc. to make it easier to use. The mobile app supports versions 8.0 or higher of the Android operating system, which means that approximately 95.4% of Android device users will be able to use the app according to the development environment Android Studio. After the tests are performed, we can assure that the system is able to solve the main use case of the application, which is the detection of SMS smishing suspects. If the received message is suspected of smishing, the user is alerted by a notification that the application itself sends to his mobile device. In addition, the user can keep a check on SMS suspected of smishing thanks to the possibility of obtaining the record of smishing or viewing analysis statistics. This project is a small contribution and the first approach to solving this serious problem. There is a lot of future work to refine this application, from the use of AI models that make the accuracy of analysis greater, to the evolution of the application to combat other types of phishing. This could significantly reduce the number of cases where older people are victims of virtual scams, making their experience with using new technologies safer and more rewarding.

   @phdthesis{sanz_barrio_diseno_2024,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Diseño de una aplicación móvil para la detección de smishing dirigida a personas mayores},
     url = {https://oa.upm.es/90970/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {Sanz Barrio, Rubén},
     month = dec,
     year = {2024},
     keywords = {keywords: Smishing, Digital Security, Older Adults, Machine Learning, Mobile Application, pdf: https://oa.upm.es/90970/1/PFG\_RUBEN\_SANZ\_BARRIO.pdf, sdg-10, sdg-16, supervisor: Lucas-Martínez, Néstor},
     supervisor = {Lucas-Martínez, Néstor},
     dimensions = {true}
   }

2. Diseño y desarrollo de un sistema de apuntamiento estabilizado en 3 ejes basado en microprocesadores para la obtención de imágenes auto niveladas y orientadas magnéticamente = Design and development of a three-axis stabilized pointing system based on microprocessors for obtaining self-leveled and magnetically oriented images

   César García Moreno

   Supervisor: Néstor Lucas-Martínez

   Jul 2024

   Abs Bib PDF Website

   The purpose of this project, titled "Design and Development of a 3-Axis Stabilized Pointing System Based on Microprocessors for Obtaining Auto-Leveled and Magnetically Oriented Images," is to design, implement, manufacture, and validate the necessary electronics to enable the rapid, precise, and semi-automatic creation of antenna visibility masks. Currently, ground-satellite communication is crucial in both civilian and military fields. Its constant use is fundamental for tasks such as geolocation, obtaining meteorological, scientific, and strategic data, as well as establishing communications, among many other applications. This communication takes place during specific periods called communication windows, which are intervals when the satellite and the antenna have a direct line of sight. Due to the high speed of low-orbit satellites, which constitute most operational satellites today, these communication windows are brief, typically between 10 and 15 minutes. Therefore, it is essential that these windows are not limited by objects in the antenna’s environment. To ensure optimal communication, a visibility mask is used to evaluate the full range of visibility between the antenna and the satellite. Specifically, a visibility mask is a set of data that relates an antenna’s azimuth position to the minimum elevation necessary to maintain visibility with the satellite. To generate visibility masks in a semi-automatic, rapid, and precise manner, it is proposed to implement a system based on capturing and processing images around the antenna to obtain the azimuth and elevation values of the visibility mask. The system developed in this project allows for the remote acquisition of images that are automatically leveled and geographically oriented. To implement a valid system for this task, a set of sensors and actuators are studied and implemented to ensure the validity of the images for subsequent processing. The obtained results confirm that with current technology, a system capable of generating a set of correctly calibrated, leveled, and oriented images for creating visibility masks with an error of less than 0,5° in azimuth and elevation can be developed. The results have also identified some system limitations and improvements to overcome these limitations. In conclusion, the implemented system guarantees high precision in identifying communication windows and optimizes the time and resources necessary for the deployment and adjustment of antennas.

   @phdthesis{garcia_moreno_diseno_2024,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Diseño y desarrollo de un sistema de apuntamiento estabilizado en 3 ejes basado en microprocesadores para la obtención de imágenes auto niveladas y orientadas magnéticamente = {Design} and development of a three-axis stabilized pointing system based on microprocessors for obtaining self-leveled and magnetically oriented images},
     url = {https://oa.upm.es/88486/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {García Moreno, César},
     month = jul,
     year = {2024},
     keywords = {external\_supervisor: Montes Fernández, Jorge, keywords: Electronic Systems, Image Processing, Visibility Masks, Antennas, Self-Leveling Images, Magnetic Orientation, Ground-Satellite Communication, Pointing Systems, pdf: https://oa.upm.es/88486/1/PFG\_CESAR\_GARCIA\_MORENO.pdf, sdg-9, supervisor: Lucas-Martínez, Néstor},
     supervisor = {Lucas-Martínez, Néstor},
     dimensions = {true}
   }

3. Sistema de comunicación y procesado para la obtención de máscaras de antena

   César García Moreno

   Supervisor: Néstor Lucas-Martínez

   Jul 2024

   Abs Bib PDF Website

   The purpose of this project, titled "Communication and processing system for obtaining horizon elevation" is the design and implementation of a system architecture to create horizon elevation reports, allowing the control of multiple hardware devices and being controlled from various types of devices. Currently, ground-satellite communication is crucial in both civilian and military fields. Its constant use is fundamental for tasks such as geolocation, obtaining meteorological, scientific, and strategic data, as well as establishing communications, among many other applications. This communication occurs during specific periods called communication windows, which are intervals when the satellite and the antenna have a direct line of sight. Due to the high speed of low orbit satellites, which most operational satellites are, these communication windows are brief, typically between 10 and 15 minutes. Therefore, it is essential that these windows are not limited by objects in the antenna’s environment. To ensure optimal communication, a visibility mask is used, which evaluates the full range of visibility between the antenna and the satellite. Specifically, a visibility mask is a set of data that relates an azimuth position of the antenna to the minimum elevation necessary to maintain visibility with the satellite. To generate visibility masks in a semi-automatic, quick, and accurate manner, it is proposed to implement a system based on capturing and processing images around the antenna to obtain the azimuth and elevation values for the visibility mask. The system developed in this project allows the processing of a set of images whose pointing direction and camera characteristics are known, to generate visibility masks from this data. The results obtained confirm that, with current technology, it is possible to develop a system capable of transmitting, processing, and displaying visibility masks with an error of less than 0.5° in azimuth and 0.5° in elevation. While achieving the results, some limitations of the system were also identified, as well as possible improvements to overcome these limitations. In conclusion, the implemented system not only guarantees high precision in identifying communication windows but also optimizes the time and resources needed for the deployment and adjustment of antennas.

   @phdthesis{garcia_moreno_sistema_2024,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Sistema de comunicación y procesado para la obtención de máscaras de antena},
     url = {https://oa.upm.es/88178/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {García Moreno, César},
     month = jul,
     year = {2024},
     keywords = {external\_supervisor: Montes Fernández, Jorge, keywords: Visibility Masks, Ground-Satellite Communication, Communication Windows, Azimuth and Elevation, Image Processing, pdf: https://oa.upm.es/88178/1/PFG\_CESAR\_GARCIA\_MORENO.pdf, sdg-3, sdg-9, supervisor: Lucas-Martínez, Néstor},
     supervisor = {Lucas-Martínez, Néstor},
     dimensions = {true}
   }

2019

1. Simulación de planificación de acciones en vehículos aéreos no tripulados aplicados a la ganadería de precisión

   Miguel García Serrano

   Supervisor: José-Fernán Martínez-Ortega

   Apr 2019

   Abs Bib PDF Website

   The growing population increase translates into a growing demand for food that must be met. This requires seeking new techniques to generate greater food production. For decades, people have used technology to improve farming by making them more efficient, such as automatic fertilization and irrigation systems, or in livestock farms, such as automatic dairy farms, where milk is extracted from cows without human intervention. With the progress made in recent years in the area of telecommunications and with the cost reductions of certain services, hardware and software, it has become possible to use technologies in different economic sectors in which 30 years ago it was unthinkable. Such is the case of the agricultural sector, in which in recent years it has become common to use GPS technology, humidity sensors, thermal sensors or even autonomous systems for the control of crop machinery and livestock farms. On the other hand, due to the proliferation of Unmanned Aerial Vehicles (UAVs), also commonly known as drones, it has been possible to have another point of view in agriculture and livestock tasks, making it common to have this type of vehicle to develop the existing work in this sector. However, any robotic system requires the ability to plan, a capacity that for little autonomy that this robotic system can have, it is very complex to develop, and more complex even in a system that requires the coordination of devices that are in such an unfavorable environment as is the air. This is the case of the control and monitoring robotic system of a livestock farm based on UAVs that this project aims to develop. Finally, thanks to computer simulation, it is possible to test whether a robotic system, like a swarm of drones, adapts to the needs established from the beginning for a specific system, all without having to waste money testing it. With the advance of computing, these simulators are able to give more and better benefits to users, allowing reproducing faithfully the characteristics of the scenarios that are simulated.

   @phdthesis{garcia_serrano_simulacion_2019,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Simulación de planificación de acciones en vehículos aéreos no tripulados aplicados a la ganadería de precisión},
     url = {https://oa.upm.es/62815/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {García Serrano, Miguel},
     month = apr,
     year = {2019},
     keywords = {keywords: Robotics; Livestock Farming; Action Planning, pdf: https://oa.upm.es/62815/1/TFG\_MIGUEL\_GARCIA\_SERRANO.pdf, sdg-2, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

2018

1. Anticipative system for a planner through machine learning techniques

   Jorge García Paredes

   Supervisor: José-Fernán Martínez-Ortega

   Jul 2018

   Abs Bib PDF Website

   Oceans and seas exploration is a largely uncharted research field with a lot of mysteries. The reason of that is mainly the physical impossibility of human beings to survive in that unknown environment. Technological innovations in the last decades make the work in unfavourable environments more likely every day. That is done by the use of robotics vehicles, which can be manned by a human expert or be autonomous. This project development focuses on planning for autonomous underwater vehicles (AUVs). The main idea is the improvement of the plan quality, as well as supply the AUVs with more autonomy. Machine Learning techniques will be used to accomplish this task. The success of the plan could be increased at the same time that the consumption of resources —as time and energy— decreases. It is difficult to work in the underwater environment due to the unexplored surrounds and the unfavourable conditions. The complete behaviour of the environment is unpredictable, and therefore, unexpected events can happen during the execution of the plan. Those problems could lead the plan to its termination without achieving the goals. The vehicles will be provided with more anticipation to look for solutions and take decisions in those situations. The standard solutions to deal with these events are to fix the plan, to re-plan or, if the worst comes to the worst, to abort the mission. For both solutions, the vehicles should communicate with the mission supervisor and/or the mission planner. Underwater communications are unreliable and slow, due to their low-bandwidth. Therefore, those solutions to the events could spend too much time waiting for an answer. The proposed resolution helps to reduce the mission time, avoiding unnecessary delays.

   @phdthesis{garcia_paredes_anticipative_2018,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Anticipative system for a planner through machine learning techniques},
     url = {https://oa.upm.es/53110/},
     language = {English},
     school = {Universidad Politécnica de Madrid},
     author = {García Paredes, Jorge},
     month = jul,
     year = {2018},
     keywords = {pdf: https://oa.upm.es/53110/1/TFG\_JORGE\_GARCIA\_PAREDES.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

2. A ros node for processing images and providing robots with location capabilities

   Wenjun Duan

   Supervisor: José-Fernán Martínez-Ortega

   Jul 2018

   Abs Bib PDF Website

   This degree project aims to carry out self-localization algorithm based on image processing for robots. Several algorithms have been reviewed along this project and finally, one combining two of them, one processing images and other one managing distances to obstacles, has been selected and carried out and deployed in a proof-of-concept to check its feasibility and its performance. Briefly, the application that has been developed and tested in a simulated environment first lets the robot recognize in which room it is by taking pictures. Once it determines the room it is in, it downloads the corresponding map and determines its exact position in the room by measuring distances to obstacles. The features of images are extracted with Scale Invariant Feature Transform (SIFT) and described with highly distinctive multi-dimensional vector making features invariant to changes in illumination, scale and noise. The use of this algorithm will let the robot know the room it is in. Afterwards, the robot will use Adaptive Monte Carlo Localization (AMCL) to finish accurate localization. By combining SIFT and AMCL, the efficiency of localization algorithm improves a lot. The SIFT and AMCL algorithms are run on Robot Operating System (ROS) and the robot and its localization process is simulated in simulator Gazebo. Apart from that, considering the powering data processing capability of Matlab, the algorithms and the final application have been developed using Matlab, that provides a ROS based library to control the robot using Gazebo and process data from ROS. According to the results in later chapter, both local and global localization are successful accomplished with these algorithms.

   @phdthesis{duan_ros_2018,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {A ros node for processing images and providing robots with location capabilities},
     url = {https://oa.upm.es/53127/},
     language = {English},
     school = {Universidad Politécnica de Madrid},
     author = {Duan, Wenjun},
     month = jul,
     year = {2018},
     keywords = {pdf: https://oa.upm.es/53127/1/TFG\_WENJUN\_DUAN.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

3. Integración de la industria 4.0 en abas ERP

   Jesús Moreno Belinchón

   Supervisor: José-Fernán Martínez-Ortega

   Jul 2018

   Abs Bib PDF Website

   This project has had the collaboration of abas Ibérica, partner of abas ERP for the Iberian Peninsula. As a company, abas ERP is a comprehensive ERP software for companies in the industrial sector that helps to manage the different areas of the company in an optimal way. The main motivation has been none other than the development of a new and cutting-edge technological and commercial application, with a wide scope of implementation, in which the knowledge acquired in the Degree in Telematic Engineering can be applied both directly and indirectly. Improving one of the processes of a company in the industrial sector, such as the preparation of an order or removal of material in an agile manner for subsequent delivery to a warehouse, avoiding possible human errors through technology. Also show the possibilities of growth in this field, Industry 4.0 and augmented reality. To do this, following the principles of Industry 4.0 and the hand of augmented reality, using as an Epson BT - 300 device, as a source of data we have abas ERP, will be key to the management of information and business processes. Through an interface based on the RESTful web services architecture, called abas middleware, we will be able to establish two-way communication between smart glasses and abas ERP. One of the key points for user acceptance will be usability, for this purpose, we implement Augumenta SDK, which will allow us to interact with the application through manual gestures. To identify correctly and quickly the different real-world objects, we use QR codes, codes with a great deal of information and widely used in any sector. Nowadays, in closed, indoor environments, we have the problem of location, it is well known that GPS is not compatible with these environments. To do this, we will use Beacons, low consumption bluetooth devices, which will allow us to establish a communication with the intelligent glasses so that we are able to calculate the location in real time of the device, which will be carried by one person, indoors. With the research carried out and the subsequent development, both on a theoretical and practical level, we have created this application, capable of communicating with other systems and optimizing material removal.

   @phdthesis{moreno_belinchon_integracion_2018,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Integración de la industria 4.0 en abas {ERP}},
     url = {https://oa.upm.es/53061/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {Moreno Belinchón, Jesús},
     month = jul,
     year = {2018},
     keywords = {pdf: https://oa.upm.es/53061/1/TFG\_JESUS\_MORENO\_BELINCHON.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

2016

1. Arquitectura cognitiva para conjuntos de robots cooperantes

   Pérez Sechi

   Supervisor: José-Fernán Martínez-Ortega

   Jul 2016

   Abs Bib PDF Website

   This Project has been carried out in collaboration with CITSEM (The Software Technologies Research and Multimedia Systems Center for Sustainability) during the period of time compound between October of 2015 and January of 2016. The main motivation of this research has been building a logic structure of reduced dimensions on top of which future applications could live together in a conscious system. Here I give the implementation details for the cognitive architecture AROS (Arduino Robot Operating System) applied to the cooperative robot, Robinson. This is based in a queued communication system in other to solve problems like logical to physical address translation, process coordination or resource assignment. Alongside the cognitive architecture, it has been used and created the ATP (AROS Transport Protocol) protocol for communications, which supports the transport of message between the applications belonging to the architecture. In the protocol description that I wrote in this memory, I explain why I choose its design and the performance restrictions we have in the communications this protocol is aimed. Like any theoretical development, I required a practical demonstration of the application for the architecture as much as for the communication protocol. For this demonstration several peripheral component drivers and the applications that make use of them have been developed. One example for those peripheral components is the wireless device used as physical transmitter and receiver. I must give thanks to the open access hardware platform Arduino and to its Arduino Robot insomuch as I have been able to obtain results in a fast way making use of the C++ programming language as the logic support. This made possible to have available programming methodologies such as Queue-Base Load Leveling Pattern which is wide used in high demand systems. Hence, the culmination of this research leads to the evolution and improvement of this first version of AROS to be used in cooperative device networks where the limitations of performance, poor communications or storage are features of the working environment.

   @phdthesis{perez_sechi_arquitectura_2016,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Arquitectura cognitiva para conjuntos de robots cooperantes},
     url = {https://oa.upm.es/44527/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {Sechi, Pérez},
     month = jul,
     year = {2016},
     keywords = {pdf: https://oa.upm.es/44527/3/TFG\_CARLOS\_IGNACIO\_PEREZ\_SECHI.pdf, sdg-9, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

2. Perception for networked robot systems

   Chaoyuan Zuo

   Supervisor: José-Fernán Martínez-Ortega

   May 2016

   Abs Bib PDF Website

   With the unprecedented development of sensors and networks, the distributed robot systems can work cooperatively and explore the environment in a totally different way. By using different perception tools, a networked robot system can interact with an environment to complete specific user’s defined tasks. It is able to get abundant information from the outside world via numerous sensors, which is referred to as context, and make judgements based on perception. The graduation project focuses on the perceptional system of networked robot systems, conducting research mainly on object identification tasks. The project aims at achieving identification of furniture in an indoor environment, as well as exploring the application of a context awareness method in robotic perception systems. We present a context-aware framework for network robot systems in indoor environments, consisting of context acquisition, its process, application, and the communication layer. In this framework, the decision-tree-based context reasoning algorithm is introduced to fulfil the task of object identification. Then, we describe two crucial contexts in the system: location and size of an object, both of which play a vital role in object identification. We first describe the map building algorithm to obtain the location information. Then, by analyzing the existing problems for robot map building and sensing limitations, we present the grid map building method based on ultrasonic sensors. Furthermore, we discuss the object detection algorithms for image processing. One visual recognition flow is introduced for robotic systems too. Finally, we take Arduino as the platform for the experiments and we verify implementation of the system in the last part of this manuscript.

   @phdthesis{zuo_perception_2016,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Perception for networked robot systems},
     url = {https://oa.upm.es/44487/},
     language = {English},
     school = {Universidad Politécnica de Madrid},
     author = {Zuo, Chaoyuan},
     month = may,
     year = {2016},
     keywords = {pdf: https://oa.upm.es/44487/1/TFG\_CHAOYUAN\_ZUO.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

2013

1. Open Technologies for prototyping the Internet of Things

   María Jesús Martínez Abascal

   Supervisor: José-Fernán Martínez-Ortega

   Sep 2013

   Abs Bib PDF Website

   The worldwide "hyper-connection" of any object around us is the challenge that promises to cover the paradigm of the Internet of Things. If the Internet has colonized the daily life of more than 2000 million1 people around the globe, the Internet of Things faces of connecting more than 100000 million2 "things" by 2020. The underlying Internet of Things’ technologies are the cornerstone that promises to solve interrelated global problems such as exponential population growth, energy management in cities, and environmental sustainability in the average and long term. On the one hand, this Project has the goal of knowledge acquisition about prototyping technologies available in the market for the Internet of Things. On the other hand, the Project focuses on the development of a system for devices management within a Wireless Sensor and Actuator Network to offer some services accessible from the Internet. To accomplish the objectives, the Project will begin with a detailed analysis of various “open source” hardware platforms to encourage creative development of applications, and automatically extract information from the environment around them for transmission to external systems. In addition, web platforms that enable mass storage with the philosophy of the Internet of Things will be studied. The project will culminate in the proposal and specification of a service-oriented software architecture for embedded systems that allows communication between devices on the network, and the data transmission to external systems. Furthermore, it abstracts the complexities of hardware to application developers.

   @phdthesis{martinez_abascal_open_2013,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Open {Technologies} for prototyping the {Internet} of {Things}},
     url = {https://oa.upm.es/22441/},
     language = {English},
     school = {Universidad Politécnica de Madrid},
     author = {Martínez Abascal, María Jesús},
     month = sep,
     year = {2013},
     keywords = {pdf: https://oa.upm.es/22441/1/PFC\_MARIA\_JESUS\_MARTINEZ\_ABASCAL.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

2. Virtualización de dispositivos para servicios en la Ciudad del Futuro

   Néstor Lucas-Martínez

   Supervisor: José-Fernán Martínez-Ortega

   Sep 2013

   Abs Bib PDF Website

   The increasing of the capabilities of all kind of devices is causing a revolution in the field of the provision of services, both in quantity and in diversity. This situation has highlighted the need to address unprecedented technological development, where the forecast of interconnected and interoperable devices between them and human beings reaches the order of billions. And these numbers go further when the connectivity of constrained networks is taken into account. This idea of an interconnected world of things has led to a vision that has been called "The Internet of Things". It’s a vision of a world where things of any kind can interact with other things, even those in the domain of a constrained network. This also leads to the creation of new composed services that exceed the sum of the parts. Besides the technological interest, this new vision relates with the one from the Smart City. A concept that uses the convergence of the energy, the transport, and the information and communication technologies to define a way to achieve sustainable and competitive growth, improving the quality of life, and opening the governance of the cities to the participation. In the development pathway to reach these goals, this Final Degree Dissertation proposes a way for the virtualization of the services offered by the variety of devices that are reaching the ability to interoperate in a network. For this it is supported by a service oriented middleware called nSOM that has been developed at EUITT. Using this architecture the goals proposed for this project are the design and development of a service gateway that makes available the resources of a sensor network through a web interface; the design and development of a Device & Service Registry according to the reference architecture proposal for the Internet of Things; and the study and design of a composition framework for orchestrated services in constrained networks. To achieve these goals this dissertation begins with a State of the Art study where the background knowledge about the technologies in use for the interoperation of things will be settled. At first it starts talking about Wireless Sensor and Actuator Networks, the architectures for Machine-to-Machine communication and Internet of Things, and also the concepts for the Web of Things vision. Next the related network and services technologies are explored, ending with a brief review of service composition technologies. Then will follow a detailed description of the nSOM architecture, and also of the proposed design for this project. Finally a scenario will be proposed where a series of validation tests will be conducted.

   @phdthesis{lucas-martinez_virtualizacion_2013,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Virtualización de dispositivos para servicios en la {Ciudad} del {Futuro}},
     url = {https://oa.upm.es/22159/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {Lucas-Martínez, Néstor},
     month = sep,
     year = {2013},
     keywords = {pdf: https://oa.upm.es/22159/1/PFC\_NESTOR\_LUCAS\_MARTINEZ.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

3. Sistema de Aparcamiento Inteligente Aplicado a las Smart Cities

   David Herrador Muñoz

   Supervisor: José-Fernán Martínez-Ortega

   Jul 2013

   Abs Bib PDF Website

   In the last decades the world has experienced an exponential increase in the use of technological solutions, which has resulted in the need to measure situations or states of the objects around us. Often, wired sensors cannot be used at many situations, so the increase in the use of technological solutions, has been translated into a increase of the need of using wireless sensors to make correct telemetries. At the social level, the increase in global demographics is closely linked to the increased need for technological services, so it is logical that more people, more technology will be consumed. The objective of this Final Project is based on the use of various nodes or so-called motes, capable of performing data transfer in wireless mode, thereby allowing performing a real application solving problems generated by the increase of population densities. Specifically looking for the realization of a smart outdoor parking system, thus helping to vehicular management tasks within the framework of the Smart Cities. The system is based on the communication protocol 802.15.4 (ZigBee) whose main characteristics lie in the low energy consumption associated to the hardware components. First there will be a State of the Art of Wireless Sensor Networks, addressing both architecture and finally the Zigbee standard XBee components to be used in this project. Then the necessary algorithms will be developed for the proper working of the intelligent parking system and finally there will be a pilot demonstrator validating the whole system.

   @phdthesis{herrador_munoz_sistema_2013,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Sistema de {Aparcamiento} {Inteligente} {Aplicado} a las {Smart} {Cities}},
     url = {https://oa.upm.es/21414/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {Herrador Muñoz, David},
     month = jul,
     year = {2013},
     keywords = {pdf: https://oa.upm.es/21414/1/PFC\_DAVID\_HERRADOR\_MU\%C3\%91OZ.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

2012

1. Redes y servicios ubicuos para Internet de las Cosas en dispositivos llevables

   Alexandra Cuerva García

   Supervisor: José-Fernán Martínez-Ortega

   Sep 2012

   Abs Bib PDF Website

   In last decades there have been significant advances in technology, which has resulted in important growth of Wireless Sensor Networks (WSN). These networks consist of a small set of nodes, also known as spots; equipped with various types of sensors. Wireless Sensor Networks can be very useful in environments where deployment of wired networks, formed by computers, routers or other network devices is not possible. However, these networks have a number of shortcomings or challenges to, sometimes, their implementation and deployment. The main objectives of this Final Project are to design and implement an agent that makes use of Bluetooth technology so you can communicate with both the service-oriented architecture, via radio, as with Bioharness module for physiological parameters; offer simple services to Wireless Sensor Network, designing an algorithm for an alarm system, make and implement a gateway between protocols using the standard IEEE802.15.4 (ZigBee) and IEEE802.15.1 standard Bluetooth Technology. Finally, implement an Android application for WiMM watch that can receive real-time alerts through the Bluetooth interface. In order to achieve these objectives, firstly we are going to carry out a study of the State of the Art in Wireless Sensor Network, where we study the architecture, the Bluetooth standard and Bluetooth devices that have been used in this project. Then, we will describe in detail the iWRAP firmware version 4, focusing on their operation modes, AT commands and errors that may occur. Therefore, we will describe the architecture and specification nSOM, to enter into the service-oriented architecture. Finally, we will execute the phase of validation of the system in a real application scenario, analyzing the results obtained during the testing phase.

   @phdthesis{cuerva_garcia_redes_2012,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Redes y servicios ubicuos para {Internet} de las {Cosas} en dispositivos llevables},
     url = {https://oa.upm.es/13950/},
     language = {Spanish},
     school = {Universidad Politécnica de Madrid},
     author = {Cuerva García, Alexandra},
     month = sep,
     year = {2012},
     keywords = {pdf: https://oa.upm.es/13950/1/PFC\_ALEXANDRA\_CUERVA\_GARCIA.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }

2. Semantic middleware development for the Internet of Things

   Jesús Rodríguez Molina

   Supervisor: José-Fernán Martínez-Ortega

   Sep 2012

   Abs Bib PDF Website

   After the extraordinary spread of the World Wide Web during the last fifteen years, engineers and developers are pushing now the Internet to its next border. A new conception in computer science and networks communication has been burgeoning during roughly the last decade: a world where most of the computers of the future will be extremely downsized, to the point that they will look like dust at its most advanced prototypes. In this vision, every single element of our “real” world has an intelligent tag that carries all their relevant data, effectively mapping the “real” world into a “virtual” one, where all the electronically augmented objects are present, can interact among them and influence with their behaviour that of the other objects, or even the behaviour of a final human user. This is the vision of the Internet of the Future, which also draws ideas of several novel tendencies in computer science and networking, as pervasive computing and the Internet of Things. As it has happened before, materializing a new paradigm that changes the way entities interrelate in this new environment has proved to be a goal full of challenges in the way. Right now the situation is exciting, with a plethora of new developments, proposals and models sprouting every time, often in an uncoordinated, decentralised manner away from any standardization, resembling somehow the status quo of the first developments of advanced computer networking, back in the 60s and the 70s. Usually, a system designed after the Internet of the Future will consist of one or several final user devices attached to these final users, a network –often a Wireless Sensor Network- charged with the task of collecting data for the final user devices, and sometimes a base station sending the data for its further processing to less hardware-constrained computers. When implementing a system designed with the Internet of the Future as a pattern, issues, and more specifically, limitations, that must be faced are numerous: lack of standards for platforms and protocols, processing bottlenecks, low battery lifetime, etc. One of the main objectives of this project is presenting a functional model of how a system based on the paradigms linked to the Internet of the Future works, overcoming some of the difficulties that can be expected and showing a model for a middleware architecture specifically designed for a pervasive, ubiquitous system. This Final Degree Dissertation is divided into several parts. Beginning with an Introduction to the main topics and concepts of this new model, a State of the Art is offered so as to provide a technological background. After that, an example of a semantic and service-oriented middleware is shown; later, a system built by means of this semantic and service-oriented middleware, and other components, is developed, justifying its placement in a particular scenario, describing it and analysing the data obtained from it. Finally, the conclusions inferred from this system and future works that would be good to be tackled are mentioned as well.

   @phdthesis{rodriguez_molina_semantic_2012,
     address = {Madrid, Spain},
     type = {Bachelor {Thesis}},
     title = {Semantic middleware development for the {Internet} of {Things}},
     url = {https://oa.upm.es/13841/},
     language = {English},
     school = {Universidad Politécnica de Madrid},
     author = {Rodríguez Molina, Jesús},
     month = sep,
     year = {2012},
     keywords = {pdf: https://oa.upm.es/13841/1/PFC\_JESUS\_RODRIGUEZ\_MOLINA.pdf, supervisor: Martínez-Ortega, José-Fernán},
     supervisor = {Martínez-Ortega, José-Fernán},
     dimensions = {true}
   }