Tesis Telecomunicaciones

Permanent URI for this collectionhttp://repositorio.uta.edu.ec/handle/123456789/34848

Browse

Filters

20241

Settings

Author: search.filters.author.Altamirano Tixe, Diego Patricio

Search Results

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Sistema de control para la optimización de trayectorias en plataformas móviles mediante computadoras industriales (IPCS) y algoritmos de aprendizaje profundo
    (Universidad Técnica de Ambato. Facultad de Ingeniería en Sistemas, Electrónica e Industrial. Carrera de Telecomunicaciones, 2024-08) Altamirano Tixe, Diego Patricio; Manzano Villafuerte, Víctor Santiago
    The growth in global demand for industrial robots is revolutionizing the manufacturing industry, underscoring the need to develop efficient and safe autonomous navigation systems. Traditional methods of trajectory optimization have difficulty adapting to unforeseen changes in dynamic environments, these problems reduce operational efficiency in the industrial industry. This project seeks to optimize the trajectories of the omnidirectional mobile platform of the KUKA youBot robot using an industrial computer (IPC) and deep learning algorithms. These algorithms allow the robot to learn and generalize movement patterns by optimizing trajectories in real time. This system consists of two nodes, the KUKA youBot robot as the master node and the IPC as the publisher node. To configure communication between nodes, the TCP/IP protocol and ROS functionalities, such as rosmaster URIs and ROS_IP, are used. In the master node, it performs data sampling with a LIDAR sensor and executes the generated trajectories, while the publisher node selects the environment and performs the execution of the deep learning algorithms. In system testing, the DQN algorithm excelled in static scenarios with high peaks in Q-values. However, the Dueling DQN algorithm showed greater robustness and long-term stability, although it required more time and training episodes. In dynamic environments, Dueling DQN earned better rewards, excelling in situations with constant and variable changes. The efficient communication between the robot and the IPC reached an efficiency of 96.92% allowing accurate coordination in real time.