Banda inteligente para captar señales cinemáticas y electromiográficas orientada al control de movimientos

Abstract

CONDITION FOR PUBLICATION OF PROJECT. This project develops an intelligent wearable band for capturing electromyographic and kinematic signals, specifically aimed at individuals with partial upper-limb amputation resulting from occupational accidents. Its purpose is to facilitate workplace reintegration and personal autonomy through embedded artificial intelligence, enabling the classification of residual muscle movements. During the development stage, an ergonomic wearable device was designed and implemented, integrating inertial and myoelectric sensors under a modular electronic architecture. The signals were processed using windowing techniques, and two learning models were comparatively evaluated: a convolutional neural network and a feature extraction–based architecture. Experimental results demonstrated that the feature extraction model provided greater stability, successfully controlling four discrete states in a robotic hand in both simulation and physical implementations. Additionally, the mechanical design ensured proper fixation, although energy autonomy was limited by the device’s dimensions. Finally, it is concluded that the proposed solution is viable for interpreting movement intention in real time, validating its use as an assistive tool for occupational continuity. Keywords: Embedded artificial intelligence, Inertial sensors, Partial amputation, Occupational reintegration, Signal processing.