Controlo e Validação de Movimentos de um Robô Virtual

• Autor(a) principal: Noites, Gonçalo da Costa Dias
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.22/24026
• topic: ROS; Gazebo; MoveIt; NAO; IMU; movement detection; physical therapy; Domínio/Área Científica::Engenharia e Tecnologia

Physical therapy, a critical and evolving field of medicine, is crucial in improving the physical health and functional ability of people of all ages and medical prob lems. It has changed over time from a field largely concerned with rehabilitation to a multifaceted healthcare profession. Synergizing physical therapy practices with simulation technology enables the creation of virtual models that are hardware inde pendent, revolutionizing the delivery and effectiveness of physical therapy services, such as remote rehabilitation. Based on this principle, the objectives of this thesis are to develop the control of a virtualized humanoid robot, develop a mechanism for validating the robot’s movements against a baseline of rehabilitation exercises and implement a set of exercises and a set of incorrect movements to test the previous objectives. Initially, the work involved the study of human movement recognition, some of the possible sensors used in this detection and, after that, the study of the simulation platforms that can be used. As a result, it was used the middleware ROS, the Gazebo simulator, the MoveIt trajectory planning framework and the NAO robot simula tion model. Starting with the simulation and control of the robot through MoveIt, the subset of physical therapy exercises were subsequently selected and replicated in Gazebo. Afterwards, the creation of the baseline to validate the robot’s move ments and the implementation of exercises considered correct and incorrect to test it. There were selected four different exercises, being created four variations of ac cepted movements and six variations of incorrect movements. The stated objectives were successfully achieved, as the system is capable of detecting and distinguishing correct movements from incorrect ones.