Increasing Control of a Prosthetic Limb.
CLIENT. Center of Bionic Medicine, Shirley Ryan Ability Lab
PROJECT. Researching the Effect of a 'Gate' on the Myoelectric Control of a Virtual Reality (VR) Limb
ROLE. Developed and executed a research plan, presented findings via an oral presentation
CAPS Target Achievement Control (TAC) Test. Courtesy of Shirley Ryan Ability Lab.
All people rely on their arms every day and need to trust them to not drop or break items. Those who use prosthetic arms have increased difficulty in these precise movements that able-bodied users take for granted every single day. It is important for prosthetic arm users to move to the correct, intended arm position on the first try with limited mistakes. Myoelectric prosthetics use a user's muscle contractions and pattern recognition to control movement to various positions or degrees of freedom (DOF).The average classification accuracy was 97.2 ± 2.0 percent across 1-DOF classifiers and 94.1 ± 3.1 percent across 3-DOF classifiers. A gate is an additional classifier whose sole job is to detect if motion should be induced or not rather than identify a specific DOF. This gate needed to be tested to see if it led to better and more precise control of a myoelectric controlled, prosthetic arm.
Using an in-house coding program known as CAPS that coordinates python and a virtual reality setting, I developed the gate to detect motion by analyzing the EMG data for an amplitude threshold. Once this threshold was met, if the gate was on, the system would then classify which DOF was intended. I tested the additional use of a gate classifier on both able-bodied users and amputees. All subjects trained the VR system using CAPs and then used a target-achievement control (TAC) test that worked by displaying an image in the VR setting and asked the subject to move their VR arm to that position. Through an experimental set-up that tested the control (no gate) and variable (gate), I was able to uncover that the gate does lead to increased control. However, my sample size was not large enough to prove this claim with significance.
Motion schematics of VR prosthetic limb under normal conditions (left) and with 'gate' (right)
Download the final presentation here.