On the reduction of complexity problem on driving of human hand prosthesis

In this paper, a statistical analysis had been carried out on the measured joint angles of the human hand´s fingers while performing some common tasks. By exploiting the correlation existing on some couples of joints, we can reduce the number of myoelectric sensors necessary to drive a (virtual or real) hand prosthesis, while still maintaining an acceptable hand´s Degree of Freedom (DoF). In order to do such kind of analysis we measured the common hand tasks by means of our HITEG data glove we developed. The results of this analysis shows that the number of sensors can be halved, extrapolating the value of remaining sensors by means of linear regression, with an error which for most applications can result acceptable. This method will allow the subject, who has to drive an hand prosthesis, to perform all common hand actions and gestures with only few, not severe, limitations.