A multifunction prosthesis controller based on fuzzy-logic techniques

We are developing a new multifunctional prosthetic hand mechanism that will be interfaced to the user using a four site myoelectric (EMG) controller based on fuzzy-logic techniques. Three to four independent EMG sites (the maximum number that could be isolated without the cross-talk becoming unacceptable) are used to control three or four degrees-of-freedom (DOF) in the prosthetic hand. Fuzzy logic techniques are used to detect EMG onset and classify user intent. Membership functions for each EMG channel are based on EMG signals acquired during a training session. Standard clinical techniques are used to locate the four sites. Rules recognizing the different EMG levels associated with a particular function are automatically generated based on the recorded EMG training data. Results of experiments comparing this controller with other control algorithms used in prosthetics control are presented. As currently implemented this controller is capable of providing seamless sequential control i.e. sequential control without the intermediate switching steps with an update rate of 50 msec.