On-line Synergy Identification for Personalized Active Arm Prosthesis: a Feasibility Study

The number of degrees of freedom in prosthetic devices today is greater than the number of available electromyographic signals from the residual limb. This potentially causes complex and unintuitive interfaces, which in turn have been cited as a leading cause of prosthetic device abandonment. Synergistic arm prosthesis control allows for an intuitive way to provide additional information to command the motion of the prosthesis by coordinating the motion relationship between the prosthesis and the residual limb. There is a challenge in identifying effective synergies, especially as the motion of an amputee generally differs from the synergy of the typical ablebodied subjects, from person to person, and along the time of prosthesis usage. In this paper, a framework for on-line data driven optimization is proposed to identify the optimal synergy setting for an individual performing a specific task, formulated as an optimization problem based on human motor control. This is done through the characterization of the task by a cost function and the parametrization of the synergy. The proposed framework is able to characterize and identify the synergy of the task of reaching for a given location, with a healthy subject in the loop. The framework is used to drive the motion of an elbow prosthesis using its synergy with the residual shoulder movement of the human subject as input. A simple parametrization for the synergy is used in this paper to demonstrate the idea of the proposed framework. Human-in-the-loop simulation results are presented, showing the feasibility of data driven optimization for on-line synergy identification in the context of an arm prosthesis.