Human intention recognition is becoming a key part of powered prosthetics research. With the advent of smart materials, the usefulness of powered prosthetics has increased. Correspondingly, there is a greater need for control technology. Electromyography (EMG) has previously been used to control myoelectric hands; however the approach to electrode placement has been speculative at best. Carpi, Raspopovic and De Rossi have shown that dielectric elastomer actuators (DEAs) can be controlled by a variety of human electrophysiological signals, including EMG. To control a DEA device with multiple degrees of freedom using EMG, multiple electrode sites are required. This paper presents an approach to control an array of DEAs using a series of electrodes and an optimized electrode data filtering scheme to maximize classification accuracy when differentiating between hand grasps. A silicon mould of a human forearm was created with an array of electrodes embedded within it. Data from each electrode site was recorded using the Universal Electrophysiological Mapping (UnEmap) system developed at the University of Auckland Bioengineering Institute for the amplification and filtering of multiple biopotential signals. The recorded data was then processed off-line, in order to calculate spatial gradients; this would determine which electrode sites would give the best bipolar readings. The spatial gradients were then compared to each other in order to find the optimal electrode sites. Several points in the extensor compartment of the forearm were found to be useful in recognizing grasping, while several points in the flexor compartment of the forearm were found to be useful in differentiating between grasps.

Date Published: 6 April 2009
PDF: 12 pages
Proc. SPIE 7287, Electroactive Polymer Actuators and Devices (EAPAD) 2009, 728724 (6 April 2009); doi: 10.1117/12.815833

Published in SPIE Proceedings Vol. 7287:
Electroactive Polymer Actuators and Devices (EAPAD) 2009
Yoseph Bar-Cohen; Thomas Wallmersperger, Editor(s)