Operación remota y control adaptativo de un manipulador virtual de 3 grados de libertad usando 2 brazaletes MYO y reconocimiento de gestos de la mano

Abstract

A system to drive a virtual manipulator with 3 degrees of freedom in a simulated computational environment is developed in this work. The control interface uses 2 MYO bracelets from the Thalmic Labs manufacturer. The MYO bracelet gives the activity of the skeletal muscles (EMG) and data from the inertial measurement unit (IMU). Regarding to the trajectory control, two controllers have been considered: the CTC model (Calculated Torque Control) to eliminate the nonlinearities of the system when the parameters are well known. The other control used is AIC (Adaptive Inertia Control) that provides greater robustness adapting to mass changes in the final effector robot. EMG signals have been processed and it has been possible to classify 6 gestures which are used to control the virtual arm. The system is tested with 20 people who do not belong to the database of training people. The results obtained show an accuracy of 91.6% in the gesture recognition system with a time classification less than 30ms. The overall result taking into account the assessment of the 20 users, shows that the total system in terms of usability has an acceptance of 82.91%, while in functionality the system has an acceptance of 64.79%.