Identificación de la intencionalidad de movimientos del miembro superior mediante el uso de sensores de fuerza y sensores inerciales

In the field of rehabilitation of people with disabilities due to damage or lack of any of their members, the use of robotic prostheses is of great importance. At present, these prostheses present a high technological development at the level of their hardware and software, however, their operation depends directly on the human machine interface (HMI), which is in charge of recording and identifying the data generated before the intention of the movement of the member; Proper registration will allow optimal control of the prosthesis. A system was implemented that allows the identification of the intentionality of the upper limb movement using alternative techniques such as the use of force sensors and inertial motion sensors. As a first step, a bracelet was manufactured whose function was to record data pertinent to changes in pressure and position of the arm. Subsequently, the information recording protocol was carried out where parameters were established to achieve communication between the microcontroller bracelet, computer microcontroller. Through the Matlab tool, algorithms were implemented to capture the signals from the bracelet, which was used in 10 test subjects, each one of the volunteers maintained a fixed arm position and a grip at the time of registration. The registration of eight combinations was specified: two positions and four types of grip, establishing a standard procedure for the location of the bracelet, and the taking of records. All the data were processed and stored in the computer, to later be used in the construction of training sets and validation sets. Multiple validation tests were performed for two supervised learning algorithms: logistic regression and vector support machines, varying the measurement parameters such as: subject, number of trials, number of positions, inclusion and exclusion of inertial motion sensors. The results of the tests show a high index of accuracy in the prediction of the position and grip before the intention of movement of the test subject in Off Line mode. On the other hand, an application was developed in Matlab that allows to graphically evaluate the behavior in On Line mode. However, the accuracy of the prediction decreased by as much as fifty percent.

• Autores: Forero Guerrero, Jaisson Hernando
• Tipo de recurso: Trabajo de grado de pregrado
• Fecha de publicación: 2020
• Institución: Universidad Antonio Nariño
• Repositorio: Repositorio UAN
• Idioma: spa
• OAI Identifier: oai:repositorio.uan.edu.co:123456789/2212
• Acceso en línea: http://repositorio.uan.edu.co/handle/123456789/2212
• Palabra clave: Inteligencia Artificial; Reconocimiento de Patrones; Identificación intencionalidad del movimiento; Aprendizaje Automático; Miografia de la Fuerza; Aprendizaje supervisado; Inteligencia Artificial; Reconocimiento de Patrones; Identificación intencionalidad del movimiento; Aprendizaje Automático; Miografia de la Fuerza; Aprendizaje supervisado
• Rights: openAccess
• License: Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)