Sistema semi-autônomo baseado em visão neuromórfica para controle de próteses mioelétricas

Abstract

In recent years, several studies have been looking for better ways for controlling upper-limb prostheses. However, current prosthesis control methods are still far from reaching the characteristics of the human limb, such as handling different degrees of freedom naturally and automatically. Furthermore, control systems based on the traditional Direct Control approach cannot provide many grasp types as the control complexity increases exponentially and consequently increases the required cognitive load of the user. Observing these limitations, this work presents a semi-autonomous control system, which uses a neuromorphic camera attached to the prosthesis for selecting the type of grasp and correct the angle of the prosthesis wrist according to the shape and angle of the desired object, respectively. Furthermore, the system can segment the desired object from the others contained in a single scene. Finally, two pickand-place experiments were carried out with volunteers, the first using the proposed semiautonomous control system and the second using a manual control system, based on switching movements through blocks. In addition, the NASA-TLX questionnaire was used to subjectively verify the level of workload during the experiments.