A comparison of direct and pattern recognition control for a two degree-of-freedom above elbow virtual prosthesis

Abstract

Individuals with a transhumeral amputation have a large functional deficit and require basic functions out of their prosthesis. Myoelectric prostheses have used amplitude control techniques for decades to restore one or two degrees of freedom to these patients. Pattern recognition control has also been investigated for transhumeral amputees, but in recent years, has been more focused on transradial amputees or high-level patients who have received targeted muscle reinnervation. This study seeks to use the most recent advances in pattern recognition control and investigate techniques that could be applied to the majority of the transhumeral amputee population that has not had the reinnervation surgery to determine if pattern recognition systems may provide them with improved control. In this study, able-bodied control subjects demonstrated that highly accurate two degree-of-freedom pattern recognition systems may be trained using four EMG channels. Such systems may be used to better control a prosthesis in real-time when compared to conventional amplitude control with mode switching.