| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 43.88 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Dado o crescimento promissor da área de Inteligência Artificial, ficaram disponíveis nos vários setores da nossa sociedade novas possibilidades. São inúmeras as empresas e entidades que utilizam esta área em aplicações do quotidiano, a tal ponto que impossibilita a sua deteção e reconhecimento. Também no domínio da robótica, se assiste a um incremento das tecnologias de Inteligência Artificial que tornaram robôs, outrora limitados a tarefas repetitivas em linhas de montagem, capazes de realizar tarefas complexas e dinâmicas.
Na sua essência, este projeto procura explorar um pouco deste contexto atual, de modo a aplicar os conceitos investigados no desenvolvimento e implementação de um braço robótico controlado por algoritmos de inteligência artificial, para realizar tarefas interativas. Assim, apresenta-se a conceção e a implementação de um sistema robótico integrado capaz de detetar objetos através de visão computacional, tomar decisões com
base em estratégias e lógica e executar tarefas físicas, utilizando um braço robótico. O foco principal é o desenvolvimento de um módulo inteligente capaz de realizar uma tarefa simples interativa como o jogo do galo. O sistema pode ser controlado através de uma interface gráfica que possibilita o acompanhamento em tempo real de todo o processo, facilitando a interação entre o utilizador e o braço robótico. Desde o controlo dos ângulos dos servo-motores ou da posição final do atuador-final, passando pela visualização das deteções nas imagens até à apresentação da melhor jogada para o estado do tabuleiro de jogo, esta interface proporciona um ambiente claro e compreensivo. No final, apresentam-se fundamentalmente os resultados obtidos, evidenciando o controlo do braço robótico pelas equações de cinemática, o treino de um modelo de
YOLOv5 robusto e a capacidade de aprendizagem de um algoritmo de Q-Learning.
Due to the promising growth in the field of Artificial Intelligence, new possibilities have become available in various sectors. Numerous companies and organizations are using this area in everyday applications, to such an extent that it is impossible to detect and recognize. Also, in the field of robotics, there has been an increase in Artificial Intelligence technologies that have made robots, once limited to repetitive tasks on assembly lines, capable of performing complex and dynamic tasks. In essence, this project seeks to explore some of this current context in order to apply the concepts investigated in the development and implementation of a robotic arm controlled by artificial intelligence algorithms to perform interactive tasks. Thus, the design and implementation of an integrated robotic system capable of detecting objects through computer vision, making decisions based on strategies and logic and performing physical tasks, using a robotic arm, is presented. The main focus is the development of an intelligent module capable of performing a simple interactive task such as the Tic-Tac-Toe game. The system can be controlled using a graphical interface that enables real-time monitoring of the entire process, facilitating interaction between the user and the robotic arm. From controlling the angles of the servomotors or the final position of the end-effector, to visualizing the detections in the images, to displaying the best move for the state of the game board, this interface provides a clear and comprehensive environment. At the end, the results obtained are basically presented, highlighting the control of the robotic arm by the kinematics equations, the training of a robust YOLOv5 model and the learning capacity of a Q-Learning algorithm.
Due to the promising growth in the field of Artificial Intelligence, new possibilities have become available in various sectors. Numerous companies and organizations are using this area in everyday applications, to such an extent that it is impossible to detect and recognize. Also, in the field of robotics, there has been an increase in Artificial Intelligence technologies that have made robots, once limited to repetitive tasks on assembly lines, capable of performing complex and dynamic tasks. In essence, this project seeks to explore some of this current context in order to apply the concepts investigated in the development and implementation of a robotic arm controlled by artificial intelligence algorithms to perform interactive tasks. Thus, the design and implementation of an integrated robotic system capable of detecting objects through computer vision, making decisions based on strategies and logic and performing physical tasks, using a robotic arm, is presented. The main focus is the development of an intelligent module capable of performing a simple interactive task such as the Tic-Tac-Toe game. The system can be controlled using a graphical interface that enables real-time monitoring of the entire process, facilitating interaction between the user and the robotic arm. From controlling the angles of the servomotors or the final position of the end-effector, to visualizing the detections in the images, to displaying the best move for the state of the game board, this interface provides a clear and comprehensive environment. At the end, the results obtained are basically presented, highlighting the control of the robotic arm by the kinematics equations, the training of a robust YOLOv5 model and the learning capacity of a Q-Learning algorithm.
Description
Keywords
Artificial intelligence Robotics Robotic arm Direct kinematics Inverse kinematics Denavit-Hartenberg convention Computer vision YOLO Q-Learning MiniMax Tic-Tac-Toe game Inteligência artificial Robótica Braço robótico Cinemática direta Cinemática inversa Convenção de Denavit-Hartenberg Visão computacional Jogo do Galo